linux.git
11 months agoMerge branch 'topic/kdump-hotplug' into next
Michael Ellerman [Mon, 13 May 2024 13:12:08 +0000 (23:12 +1000)]
Merge branch 'topic/kdump-hotplug' into next

Merge our topic branch containing kdump hotplug changes, more detail from the
original cover letter:

Commit 247262756121 ("crash: add generic infrastructure for crash
hotplug support") added a generic infrastructure that allows
architectures to selectively update the kdump image component during CPU
or memory add/remove events within the kernel itself.

This patch series adds crash hotplug handler for PowerPC and enable
support to update the kdump image on CPU/Memory add/remove events.

Among the 6 patches in this series, the first two patches make changes
to the generic crash hotplug handler to assist PowerPC in adding support
for this feature. The last four patches add support for this feature.

The following section outlines the problem addressed by this patch
series, along with the current solution, its shortcomings, and the
proposed resolution.

Problem:
========
Due to CPU/Memory hotplug or online/offline events the elfcorehdr
(which describes the CPUs and memory of the crashed kernel) and FDT
(Flattened Device Tree) of kdump image becomes outdated. Consequently,
attempting dump collection with an outdated elfcorehdr or FDT can lead
to failed or inaccurate dump collection.

Going forward CPU hotplug or online/offline events are referred as
CPU/Memory add/remove events.

Existing solution and its shortcoming:
======================================
The current solution to address the above issue involves monitoring the
CPU/memory add/remove events in userspace using udev rules and whenever
there are changes in CPU and memory resources, the entire kdump image
is loaded again. The kdump image includes kernel, initrd, elfcorehdr,
FDT, purgatory. Given that only elfcorehdr and FDT get outdated due to
CPU/Memory add/remove events, reloading the entire kdump image is
inefficient. More importantly, kdump remains inactive for a substantial
amount of time until the kdump reload completes.

Proposed solution:
==================
Instead of initiating a full kdump image reload from userspace on
CPU/Memory hotplug and online/offline events, the proposed solution aims
to update only the necessary kdump image component within the kernel
itself.

11 months agoMerge branch 'topic/ppc-kvm' into next
Michael Ellerman [Mon, 13 May 2024 13:08:32 +0000 (23:08 +1000)]
Merge branch 'topic/ppc-kvm' into next

Merge our KVM topic branch.

11 months agopowerpc/85xx: fix compile error without CONFIG_CRASH_DUMP
Hari Bathini [Fri, 10 May 2024 08:07:57 +0000 (13:37 +0530)]
powerpc/85xx: fix compile error without CONFIG_CRASH_DUMP

Since commit 5c4233cc0920 ("powerpc/kdump: Split KEXEC_CORE and
CRASH_DUMP dependency"), crashing_cpu is not available without
CONFIG_CRASH_DUMP. Fix compile error on 64-BIT 85xx owing to this
change.

Fixes: 5c4233cc0920 ("powerpc/kdump: Split KEXEC_CORE and CRASH_DUMP dependency")
Cc: stable@vger.kernel.org # v6.9+
Reported-by: Christian Zigotzky <chzigotzky@xenosoft.de>
Closes: https://lore.kernel.org/all/fa247ae4-5825-4dbe-a737-d93b7ab4d4b9@xenosoft.de/
Suggested-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240510080757.560159-1-hbathini@linux.ibm.com
11 months agopowerpc/fadump: update documentation about bootargs_append
Hari Bathini [Fri, 10 May 2024 08:21:14 +0000 (13:51 +0530)]
powerpc/fadump: update documentation about bootargs_append

Update ABI documentation about the introduction of the new sysfs
entry bootargs_append. This sysfs entry will be used to setup the
additional parameters to be passed to dump capture kernel.

Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240510082114.561163-1-hbathini@linux.ibm.com
11 months agopowerpc/fadump: pass additional parameters when fadump is active
Hari Bathini [Thu, 9 May 2024 11:57:55 +0000 (17:27 +0530)]
powerpc/fadump: pass additional parameters when fadump is active

Append the additional parameters passed/set in the dedicated parameter
area (RTAS_FADUMP_PARAM_AREA) to bootargs in fadump capture kernel.

Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240509115755.519982-4-hbathini@linux.ibm.com
11 months agopowerpc/fadump: setup additional parameters for dump capture kernel
Hari Bathini [Thu, 9 May 2024 11:57:54 +0000 (17:27 +0530)]
powerpc/fadump: setup additional parameters for dump capture kernel

For fadump case, passing additional parameters to dump capture kernel
helps in minimizing the memory footprint for it and also provides the
flexibility to disable components/modules, like hugepages, that are
hindering the boot process of the special dump capture environment.

Set up a dedicated parameter area to be passed to the capture kernel.
This area type is defined as RTAS_FADUMP_PARAM_AREA. Sysfs attribute
'/sys/kernel/fadump/bootargs_append' is exported to the userspace to
specify the additional parameters to be passed to the capture kernel

Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240509115755.519982-3-hbathini@linux.ibm.com
11 months agopowerpc/pseries/fadump: add support for multiple boot memory regions
Hari Bathini [Thu, 9 May 2024 11:57:53 +0000 (17:27 +0530)]
powerpc/pseries/fadump: add support for multiple boot memory regions

Currently, fadump on pseries assumes a single boot memory region even
though f/w supports more than one boot memory region. Add support for
more boot memory regions to make the implementation flexible for any
enhancements that introduce other region types. For this, rtas memory
structure for fadump is updated to have multiple boot memory regions
instead of just one. Additionally, methods responsible for creating
the fadump memory structure during both the first and second kernel
boot have been modified to take these multiple boot memory regions
into account. Also, a new callback has been added to the fadump_ops
structure to get the maximum boot memory regions supported by the
platform.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240509115755.519982-2-hbathini@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Fix spelling mistake "predicition" -> "prediction"
Colin Ian King [Wed, 8 May 2024 08:41:17 +0000 (09:41 +0100)]
selftests/powerpc/dexcr: Fix spelling mistake "predicition" -> "prediction"

There is a spelling mistake in the help message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240508084117.2869261-1-colin.i.king@gmail.com
11 months agoKVM: PPC: Book3S HV nestedv2: Fix an error handling path in gs_msg_ops_kvmhv_nestedv2...
Christophe JAILLET [Sun, 28 Jan 2024 11:34:25 +0000 (12:34 +0100)]
KVM: PPC: Book3S HV nestedv2: Fix an error handling path in gs_msg_ops_kvmhv_nestedv2_config_fill_info()

The return value of kvmppc_gse_put_buff_info() is not assigned to 'rc' and
'rc' is uninitialized at this point.
So the error handling can not work.

Assign the expected value to 'rc' to fix the issue.

Fixes: 19d31c5f1157 ("KVM: PPC: Add support for nestedv2 guests")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Vaibhav Jain <vaibhav@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/a7ed4cc12e0a0bbd97fac44fe6c222d1c393ec95.1706441651.git.christophe.jaillet@wanadoo.fr
11 months agoKVM: PPC: Fix documentation for ppc mmu caps
Joel Stanley [Tue, 11 Apr 2023 06:14:46 +0000 (15:44 +0930)]
KVM: PPC: Fix documentation for ppc mmu caps

The documentation mentions KVM_CAP_PPC_RADIX_MMU, but the defines in the
kvm headers spell it KVM_CAP_PPC_MMU_RADIX. Similarly with
KVM_CAP_PPC_MMU_HASH_V3.

Fixes: c92701322711 ("KVM: PPC: Book3S HV: Add userspace interfaces for POWER9 MMU")
Signed-off-by: Joel Stanley <joel@jms.id.au>
Acked-by: Paul Mackerras <paulus@ozlabs.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20230411061446.26324-1-joel@jms.id.au
11 months agoKVM: PPC: code cleanup for kvmppc_book3s_irqprio_deliver
Kunwu Chan [Thu, 25 Jan 2024 08:33:48 +0000 (16:33 +0800)]
KVM: PPC: code cleanup for kvmppc_book3s_irqprio_deliver

This part was commented from commit 2f4cf5e42d13 ("Add book3s.c")
in about 14 years before.
If there are no plans to enable this part code in the future,
we can remove this dead code.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240125083348.533883-1-chentao@kylinos.cn
11 months agoKVM: PPC: Book3S HV nestedv2: Cancel pending DEC exception
Vaibhav Jain [Mon, 15 Apr 2024 03:57:29 +0000 (09:27 +0530)]
KVM: PPC: Book3S HV nestedv2: Cancel pending DEC exception

This reverts commit 180c6b072bf3 ("KVM: PPC: Book3S HV nestedv2: Do not
cancel pending decrementer exception") [1] which prevented canceling a
pending HDEC exception for nestedv2 KVM guests. It was done to avoid
overhead of a H_GUEST_GET_STATE hcall to read the 'DEC expiry TB' register
which was higher compared to handling extra decrementer exceptions.

However recent benchmarks indicate that overhead of not handling 'DECR'
expiry for Nested KVM Guest(L2) is higher and results in much larger exits
to Pseries Host(L1) as indicated by the Unixbench-arithoh bench[2]

Metric            | Current upstream    | Revert [1]  | Difference %
========================================================================
arithoh-count (10)    | 3244831634     | 3403089673  | +04.88%
kvm_hv:kvm_guest_exit | 513558     | 152441   | -70.32%
probe:kvmppc_gsb_recv | 28060     | 28110   | +00.18%

N=1

As indicated by the data above that reverting [1] results in substantial
reduction in number of L2->L1 exits with only slight increase in number of
H_GUEST_GET_STATE hcalls to read the value of 'DEC expiry TB'. This results
in an overall ~4% improvement of arithoh[2] throughput.

[1] commit 180c6b072bf3 ("KVM: PPC: Book3S HV nestedv2: Do not cancel pending decrementer exception")
[2] https://github.com/kdlucas/byte-unixbench/

Fixes: 180c6b072bf3 ("KVM: PPC: Book3S HV nestedv2: Do not cancel pending decrementer exception")
Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240415035731.103097-1-vaibhav@linux.ibm.com
11 months agopowerpc/xmon: Check cpu id in commands "c#", "dp#" and "dx#"
Greg Kurz [Tue, 9 Mar 2021 18:11:10 +0000 (19:11 +0100)]
powerpc/xmon: Check cpu id in commands "c#", "dp#" and "dx#"

All these commands end up peeking into the PACA using the user
originated cpu id as an index. Check the cpu id is valid in order
to prevent xmon to crash. Instead of printing an error, this follows
the same behavior as the "lp s #" command : ignore the buggy cpu id
parameter and fall back to the #-less version of the command.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/161531347060.252863.10490063933688958044.stgit@bahia.lan
11 months agopowerpc/code-patching: Use dedicated memory routines for patching
Benjamin Gray [Mon, 25 Mar 2024 05:28:15 +0000 (16:28 +1100)]
powerpc/code-patching: Use dedicated memory routines for patching

The patching page set up as a writable alias may be in quadrant 0
(userspace) if the temporary mm path is used. This causes sanitiser
failures if so. Sanitiser failures also occur on the non-mm path
because the plain memset family is instrumented, and KASAN treats the
patching window as poisoned.

Introduce locally defined patch_* variants of memset that perform an
uninstrumented lower level set, as well as detecting write errors like
the original single patch variant does.

copy_to_user() is not correct here, as the PTE makes it a proper kernel
page (the EAA is privileged access only, RW). It just happens to be in
quadrant 0 because that's the hardware's mechanism for using the current
PID vs PID 0 in translations. Importantly, it's incorrect to allow user
page accesses.

Now that the patching memsets are used, we also propagate a failure up
to the caller as the single patch variant does.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240325052815.854044-2-bgray@linux.ibm.com
11 months agopowerpc/code-patching: Test patch_instructions() during boot
Benjamin Gray [Mon, 25 Mar 2024 05:28:14 +0000 (16:28 +1100)]
powerpc/code-patching: Test patch_instructions() during boot

patch_instructions() introduces new behaviour with a couple of
variations. Test each case of

  * a repeated 32-bit instruction,
  * a repeated 64-bit instruction (ppc64), and
  * a copied sequence of instructions

for both on a single page and when it crosses a page boundary.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240325052815.854044-1-bgray@linux.ibm.com
11 months agopowerpc64/kasan: Pass virtual addresses to kasan_init_phys_region()
Benjamin Gray [Mon, 12 Feb 2024 04:50:20 +0000 (15:50 +1100)]
powerpc64/kasan: Pass virtual addresses to kasan_init_phys_region()

The kasan_init_phys_region() function maps shadow pages necessary for
the ranges of the linear map backed by physical pages. Currently
kasan_init_phys_region() is being passed physical addresses, but
kasan_mem_to_shadow() expects virtual addresses.

It works right now because the lower bits (12:64) of the
kasan_mem_to_shadow() calculation are the same for the real and virtual
addresses, so the actual PTE value is the same in the end. But virtual
addresses are the intended input, so fix it.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240212045020.70364-1-bgray@linux.ibm.com
11 months agopowerpc: rename SPRN_HID2 define to SPRN_HID2_750FX
Matthias Schiffer [Wed, 24 Jan 2024 10:50:31 +0000 (11:50 +0100)]
powerpc: rename SPRN_HID2 define to SPRN_HID2_750FX

This register number is hardware-specific, rename it for clarity.

FIXME comments are added in a few places where it seems like the wrong
register is used. As I can't test this, only the rename is done with no
functional change.

Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240124105031.45734-1-matthias.schiffer@ew.tq-group.com
11 months agopowerpc: Fix typos
Bjorn Helgaas [Wed, 3 Jan 2024 23:16:04 +0000 (17:16 -0600)]
powerpc: Fix typos

Fix typos, most reported by "codespell arch/powerpc".  Only touches
comments, no code changes.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240103231605.1801364-8-helgaas@kernel.org
11 months agopowerpc/eeh: Fix spelling of the word "auxillary" and update comment
Ghanshyam Agrawal [Thu, 28 Dec 2023 10:15:18 +0000 (15:45 +0530)]
powerpc/eeh: Fix spelling of the word "auxillary" and update comment

Fix spelling of the word "auxillary" in arch/powerpc/kernel/eeh_pe.c
and arch/powerpc/include/asm/eeh.h.

Also update the eeh_set_pe_aux_size() comment to include the units.

Signed-off-by: Ghanshyam Agrawal <ghanshyam1898@gmail.com>
[mpe: Squash into one commit]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/2ab034609285b21c309cd8ab26c937c846d37ee7.1703756365.git.ghanshyam1898@gmail.com
11 months agomacintosh/ams: Fix unused variable warning
Michael Ellerman [Tue, 7 May 2024 14:01:50 +0000 (00:01 +1000)]
macintosh/ams: Fix unused variable warning

If both CONFIG_SENSORS_AMS_PMU and CONFIG_SENSORS_AMS_I2C are unset,
there is an unused variable warning in the ams driver:

  drivers/macintosh/ams/ams-core.c: In function 'ams_init':
  drivers/macintosh/ams/ams-core.c:181:29: warning: unused variable 'np'
    181 |         struct device_node *np;

The driver needs at least one of the configs enabled in order to
actually function. So fix the compiler warning by ensuring at least one
of the configs is enabled.

Suggested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240507140150.54630-1-mpe@ellerman.id.au
11 months agopowerpc/Makefile: Remove bits related to the previous use of -mcmodel=large
Naveen N Rao [Wed, 10 Jan 2024 14:12:37 +0000 (19:42 +0530)]
powerpc/Makefile: Remove bits related to the previous use of -mcmodel=large

All supported compilers today (gcc v5.1+ and clang v11+) have support for
-mcmodel=medium. As such, NO_MINIMAL_TOC is no longer being set. Remove
NO_MINIMAL_TOC as well as the fallback to -mminimal-toc.

Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Naveen N Rao <naveen@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240110141237.3179199-1-naveen@kernel.org
11 months agopowerpc/pseries/pci: Code cleanup
Kunwu Chan [Fri, 26 Jan 2024 02:50:30 +0000 (10:50 +0800)]
powerpc/pseries/pci: Code cleanup

This part was commented in about 19 years before.
If there are no plans to enable this part code in the future,
we can remove this dead code.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240126025030.577795-1-chentao@kylinos.cn
11 months agopowerpc/cell: Code cleanup for spufs_mfc_flush
Kunwu Chan [Fri, 26 Jan 2024 02:12:58 +0000 (10:12 +0800)]
powerpc/cell: Code cleanup for spufs_mfc_flush

This part was commented from commit a33a7d7309d7
("[PATCH] spufs: implement mfc access for PPE-side DMA")
in about 18 years before.

If there are no plans to enable this part code in the future,
we can remove this dead code.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Suggested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240126021258.574916-1-chentao@kylinos.cn
11 months agopowerpc/iommu: Code cleanup for cell/iommu.c
Kunwu Chan [Thu, 25 Jan 2024 08:26:37 +0000 (16:26 +0800)]
powerpc/iommu: Code cleanup for cell/iommu.c

This part was commented from commit 165785e5c0be ("[POWERPC] Cell
iommu support") in about 17 years before.

If there are no plans to enable this part code in the future,
we can remove this dead code.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240125082637.532826-1-chentao@kylinos.cn
11 months agopowerpc: Fix preserved memory size for int-vectors
GUO Zihua [Sat, 13 Jan 2024 08:05:09 +0000 (08:05 +0000)]
powerpc: Fix preserved memory size for int-vectors

The first 32k of memory is reserved for interrupt vectors, however for
powerpc64 this might not be enough. Fix this by reserving the maximum
size between 32k and the real size of interrupt vectors.

Signed-off-by: GUO Zihua <guozihua@huawei.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240113080509.1598290-1-guozihua@huawei.com
11 months agopowerpc: dts: fsl: rename ifc node name to be memory-controller
Li Yang [Fri, 19 Jan 2024 20:38:57 +0000 (15:38 -0500)]
powerpc: dts: fsl: rename ifc node name to be memory-controller

Update the node name to be align with binding document.

Signed-off-by: Li Yang <leoyang.li@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240119203911.3143928-4-Frank.Li@nxp.com
11 months agopowerpc: dts: mpc85xx: remove "simple-bus" compatible from ifc node
Li Yang [Fri, 19 Jan 2024 20:38:56 +0000 (15:38 -0500)]
powerpc: dts: mpc85xx: remove "simple-bus" compatible from ifc node

Update dts to match dts binding document.

Signed-off-by: Li Yang <leoyang.li@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240119203911.3143928-3-Frank.Li@nxp.com
11 months agopowerpc: dts: p1010rdb: fix INTx interrupt issue on P1010RDB-PB
Xiaowei Bao [Fri, 19 Jan 2024 20:38:55 +0000 (15:38 -0500)]
powerpc: dts: p1010rdb: fix INTx interrupt issue on P1010RDB-PB

Due to the INTA is shared with the active-low PHY2 interrupt on
P1010RDB-PA board, so configure P1010RDB-PA's INTA with polarity as
active-low, the P1010RDB-PB board is used separately, so configure
P1010RDB-PB's INTA with polarity as active-high.  The INTX in
P1010RDB-PB do not work because of the pcie@0 node fixup will be
overwrited by p1010si-post.dtsi file, so we move the pcie@0 node fixup
to p1010rdb-pb.dts and p1010rdb-pb_36b.dts.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Li Yang <leoyang.li@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240119203911.3143928-2-Frank.Li@nxp.com
11 months agopowerpc: dts: add power management nodes to FSL chips
Ran Wang [Fri, 19 Jan 2024 20:38:54 +0000 (15:38 -0500)]
powerpc: dts: add power management nodes to FSL chips

Enable Power Management feature on device tree, including MPC8536,
MPC8544, MPC8548, MPC8572, P1010, P1020, P1021, P1022, P2020, P2041,
P3041, T104X, T1024.

Signed-off-by: Zhao Chenhui <chenhui.zhao@freescale.com>
Signed-off-by: Ran Wang <ran.wang_1@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240119203911.3143928-1-Frank.Li@nxp.com
11 months agopowerpc/rtas: Add kernel-doc comments to smp_startup_cpu()
Yang Li [Mon, 8 Apr 2024 05:31:09 +0000 (13:31 +0800)]
powerpc/rtas: Add kernel-doc comments to smp_startup_cpu()

This commit adds kernel-doc style comments with complete parameter
descriptions for the function smp_startup_cpu().

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240408053109.96360-2-yang.lee@linux.alibaba.com
11 months agopowerpc: Fix kernel-doc comments in fsl_gtm.c
Yang Li [Mon, 8 Apr 2024 05:31:08 +0000 (13:31 +0800)]
powerpc: Fix kernel-doc comments in fsl_gtm.c

Fix some function names in kernel-doc comments.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240408053109.96360-1-yang.lee@linux.alibaba.com
11 months agopowerpc: boot: Fix kernel-doc param for partial_decompress
Yang Li [Mon, 8 Apr 2024 08:39:16 +0000 (16:39 +0800)]
powerpc: boot: Fix kernel-doc param for partial_decompress

Fix the kernel-doc annotation for the 'skip' parameter in the
partial_decompress() function by adding a missing underscore and colon.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240408083916.123369-1-yang.lee@linux.alibaba.com
11 months agopowerpc: remove unused *_syscall_64.o variables in Makefile
Masahiro Yamada [Fri, 16 Feb 2024 13:55:17 +0000 (22:55 +0900)]
powerpc: remove unused *_syscall_64.o variables in Makefile

Commit ab1a517d55b0 ("powerpc/syscall: Rename syscall_64.c into
interrupt.c") missed to update these three lines:

  GCOV_PROFILE_syscall_64.o := n
  KCOV_INSTRUMENT_syscall_64.o := n
  UBSAN_SANITIZE_syscall_64.o := n

To restore the original behavior, we could replace them with:

  GCOV_PROFILE_interrupt.o := n
  KCOV_INSTRUMENT_interrupt.o := n
  UBSAN_SANITIZE_interrupt.o := n

However, nobody has noticed the functional change in the past three
years, so they were unneeded.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240216135517.2002749-1-masahiroy@kernel.org
11 months agopowerpc/bpf/32: Fix failing test_bpf tests
Christophe Leroy [Tue, 5 Mar 2024 15:36:23 +0000 (16:36 +0100)]
powerpc/bpf/32: Fix failing test_bpf tests

Recent additions in BPF like cpu v4 instructions, test_bpf module
exhibits the following failures:

  test_bpf: #82 ALU_MOVSX | BPF_B jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #83 ALU_MOVSX | BPF_H jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #84 ALU64_MOVSX | BPF_B jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #85 ALU64_MOVSX | BPF_H jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)
  test_bpf: #86 ALU64_MOVSX | BPF_W jited:1 ret 2 != 1 (0x2 != 0x1)FAIL (1 times)

  test_bpf: #165 ALU_SDIV_X: -6 / 2 = -3 jited:1 ret 2147483645 != -3 (0x7ffffffd != 0xfffffffd)FAIL (1 times)
  test_bpf: #166 ALU_SDIV_K: -6 / 2 = -3 jited:1 ret 2147483645 != -3 (0x7ffffffd != 0xfffffffd)FAIL (1 times)

  test_bpf: #169 ALU_SMOD_X: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)
  test_bpf: #170 ALU_SMOD_K: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)

  test_bpf: #172 ALU64_SMOD_K: -7 % 2 = -1 jited:1 ret 1 != -1 (0x1 != 0xffffffff)FAIL (1 times)

  test_bpf: #313 BSWAP 16: 0x0123456789abcdef -> 0xefcd
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 301 PASS
  test_bpf: #314 BSWAP 32: 0x0123456789abcdef -> 0xefcdab89
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 555 PASS
  test_bpf: #315 BSWAP 64: 0x0123456789abcdef -> 0x67452301
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 268 PASS
  test_bpf: #316 BSWAP 64: 0x0123456789abcdef >> 32 -> 0xefcdab89
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 269 PASS
  test_bpf: #317 BSWAP 16: 0xfedcba9876543210 -> 0x1032
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 460 PASS
  test_bpf: #318 BSWAP 32: 0xfedcba9876543210 -> 0x10325476
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 320 PASS
  test_bpf: #319 BSWAP 64: 0xfedcba9876543210 -> 0x98badcfe
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 222 PASS
  test_bpf: #320 BSWAP 64: 0xfedcba9876543210 >> 32 -> 0x10325476
  eBPF filter opcode 00d7 (@2) unsupported
  jited:0 273 PASS

  test_bpf: #344 BPF_LDX_MEMSX | BPF_B
  eBPF filter opcode 0091 (@5) unsupported
  jited:0 432 PASS
  test_bpf: #345 BPF_LDX_MEMSX | BPF_H
  eBPF filter opcode 0089 (@5) unsupported
  jited:0 381 PASS
  test_bpf: #346 BPF_LDX_MEMSX | BPF_W
  eBPF filter opcode 0081 (@5) unsupported
  jited:0 505 PASS

  test_bpf: #490 JMP32_JA: Unconditional jump: if (true) return 1
  eBPF filter opcode 0006 (@1) unsupported
  jited:0 261 PASS

  test_bpf: Summary: 1040 PASSED, 10 FAILED, [924/1038 JIT'ed]

Fix them by adding missing processing.

Fixes: daabb2b098e0 ("bpf/tests: add tests for cpuv4 instructions")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/91de862dda99d170697eb79ffb478678af7e0b27.1709652689.git.christophe.leroy@csgroup.eu
11 months agopowerpc/64: Set _IO_BASE to POISON_POINTER_DELTA not 0 for CONFIG_PCI=n
Michael Ellerman [Fri, 3 May 2024 07:56:19 +0000 (17:56 +1000)]
powerpc/64: Set _IO_BASE to POISON_POINTER_DELTA not 0 for CONFIG_PCI=n

There is code that builds with calls to IO accessors even when
CONFIG_PCI=n, but the actual calls are guarded by runtime checks.

If not those calls would be faulting, because the page at virtual
address zero is (usually) not mapped into the kernel. As Arnd pointed
out, it is possible a large port value could cause the address to be
above mmap_min_addr which would then access userspace, which would be
a bug.

To avoid any such issues, set _IO_BASE to POISON_POINTER_DELTA. That
is a value chosen to point into unmapped space between the kernel and
userspace, so any access will always fault.

Note that on 32-bit POISON_POINTER_DELTA is 0, so the patch only has an
effect on 64-bit.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240503075619.394467-2-mpe@ellerman.id.au
11 months agopowerpc/io: Avoid clang null pointer arithmetic warnings
Michael Ellerman [Fri, 3 May 2024 07:56:18 +0000 (17:56 +1000)]
powerpc/io: Avoid clang null pointer arithmetic warnings

With -Wextra clang warns about pointer arithmetic using a null pointer.
When building with CONFIG_PCI=n, that triggers a warning in the IO
accessors, eg:

  In file included from linux/arch/powerpc/include/asm/io.h:672:
  linux/arch/powerpc/include/asm/io-defs.h:23:1: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
     23 | DEF_PCI_AC_RET(inb, u8, (unsigned long port), (port), pio, port)
        | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  ...
  linux/arch/powerpc/include/asm/io.h:591:53: note: expanded from macro '__do_inb'
    591 | #define __do_inb(port)          readb((PCI_IO_ADDR)_IO_BASE + port);
        |                                       ~~~~~~~~~~~~~~~~~~~~~ ^

That is because when CONFIG_PCI=n, _IO_BASE is defined as 0.

Although _IO_BASE is defined as plain 0, the cast (PCI_IO_ADDR) converts
it to void * before the addition with port happens.

Instead the addition can be done first, and then the cast. The resulting
value will be the same, but avoids the warning, and also avoids void
pointer arithmetic which is apparently non-standard.

Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org>
Closes: https://lore.kernel.org/all/CA+G9fYtEh8zmq8k8wE-8RZwW-Qr927RLTn+KqGnq1F=ptaaNsA@mail.gmail.com
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240503075619.394467-1-mpe@ellerman.id.au
11 months agopowerpc/bpf: enable kfunc call
Hari Bathini [Thu, 2 May 2024 17:32:05 +0000 (23:02 +0530)]
powerpc/bpf: enable kfunc call

Currently, bpf jit code on powerpc assumes all the bpf functions and
helpers to be part of core kernel text. This is false for kfunc case,
as function addresses may not be part of core kernel text area. So,
add support for addresses that are not within core kernel text area
too, to enable kfunc support. Emit instructions based on whether the
function address is within core kernel text address or not, to retain
optimized instruction sequence where possible.

In case of PCREL, as a bpf function that is not within core kernel
text area is likely to go out of range with relative addressing on
kernel base, use PC relative addressing. If that goes out of range,
load the full address with PPC_LI64().

With addresses that are not within core kernel text area supported,
override bpf_jit_supports_kfunc_call() to enable kfunc support. Also,
override bpf_jit_supports_far_kfunc_call() to enable 64-bit pointers,
as an address offset can be more than 32-bit long on PPC64.

Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240502173205.142794-2-hbathini@linux.ibm.com
11 months agopowerpc/64/bpf: fix tail calls for PCREL addressing
Hari Bathini [Thu, 2 May 2024 17:32:04 +0000 (23:02 +0530)]
powerpc/64/bpf: fix tail calls for PCREL addressing

With PCREL addressing, there is no kernel TOC. So, it is not setup in
prologue when PCREL addressing is used. But the number of instructions
to skip on a tail call was not adjusted accordingly. That resulted in
not so obvious failures while using tailcalls. 'tailcalls' selftest
crashed the system with the below call trace:

  bpf_test_run+0xe8/0x3cc (unreliable)
  bpf_prog_test_run_skb+0x348/0x778
  __sys_bpf+0xb04/0x2b00
  sys_bpf+0x28/0x38
  system_call_exception+0x168/0x340
  system_call_vectored_common+0x15c/0x2ec

Also, as bpf programs are always module addresses and a bpf helper in
general is a core kernel text address, using PC relative addressing
often fails with "out of range of pcrel address" error. Switch to
using kernel base for relative addressing to handle this better.

Fixes: 7e3a68be42e1 ("powerpc/64: vmlinux support building with PCREL addresing")
Cc: stable@vger.kernel.org # v6.4+
Signed-off-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240502173205.142794-1-hbathini@linux.ibm.com
11 months agoDocumentation: Fix the address of the linuxppc-dev mailing list
Stephen Rothwell [Fri, 3 May 2024 02:10:12 +0000 (12:10 +1000)]
Documentation: Fix the address of the linuxppc-dev mailing list

This list was moved many years ago.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240503121012.3ba5000b@canb.auug.org.au
11 months agoDocumentation: Document PowerPC kernel dynamic DEXCR interface
Benjamin Gray [Wed, 17 Apr 2024 11:23:25 +0000 (21:23 +1000)]
Documentation: Document PowerPC kernel dynamic DEXCR interface

Documents how to use the PR_PPC_GET_DEXCR and PR_PPC_SET_DEXCR prctl()'s
for changing a process's DEXCR or its process tree default value.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-10-bgray@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Add chdexcr utility
Benjamin Gray [Wed, 17 Apr 2024 11:23:24 +0000 (21:23 +1000)]
selftests/powerpc/dexcr: Add chdexcr utility

Adds a utility to exercise the prctl DEXCR inheritance in the shell.
Supports setting and clearing each aspect.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
[mpe: Use correct SPDX license, use execvp() for usability, print errors]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-9-bgray@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Add DEXCR config details to lsdexcr
Benjamin Gray [Wed, 17 Apr 2024 11:23:23 +0000 (21:23 +1000)]
selftests/powerpc/dexcr: Add DEXCR config details to lsdexcr

Now that the DEXCR can be configured with prctl, add a section in
lsdexcr that explains why each aspect is set the way it is.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-8-bgray@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Attempt to enable NPHIE in hashchk selftest
Benjamin Gray [Wed, 17 Apr 2024 11:23:22 +0000 (21:23 +1000)]
selftests/powerpc/dexcr: Attempt to enable NPHIE in hashchk selftest

Now that a process can control its DEXCR to some extent, make the
hashchk tests more reliable by explicitly setting the local and onexec
NPHIE aspect.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-7-bgray@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Add DEXCR prctl interface test
Benjamin Gray [Wed, 17 Apr 2024 11:23:21 +0000 (21:23 +1000)]
selftests/powerpc/dexcr: Add DEXCR prctl interface test

Some basic tests of the prctl interface of the DEXCR.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
[mpe: Add missing SPDX tag]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-6-bgray@linux.ibm.com
11 months agopowerpc/dexcr: Add DEXCR prctl interface
Benjamin Gray [Wed, 17 Apr 2024 11:23:20 +0000 (21:23 +1000)]
powerpc/dexcr: Add DEXCR prctl interface

Now that we track a DEXCR on a per-task basis, individual tasks are free
to configure it as they like.

The interface is a pair of getter/setter prctl's that work on a single
aspect at a time (multiple aspects at once is more difficult if there
are different rules applied for each aspect, now or in future). The
getter shows the current state of the process config, and the setter
allows setting/clearing the aspect.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
[mpe: Account for PR_RISCV_SET_ICACHE_FLUSH_CTX, shrink some longs lines]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-5-bgray@linux.ibm.com
11 months agopowerpc/dexcr: Reset DEXCR value across exec
Benjamin Gray [Wed, 17 Apr 2024 11:23:19 +0000 (21:23 +1000)]
powerpc/dexcr: Reset DEXCR value across exec

Inheriting the DEXCR across exec can have security and usability
concerns. If a program is compiled with hash instructions it generally
expects to run with NPHIE enabled. But if the parent process disables
NPHIE then if it's not careful it will be disabled for any children too
and the protection offered by hash checks is basically worthless.

This patch introduces a per-process reset value that new execs in a
particular process tree are initialized with. This enables fine grained
control over what DEXCR value child processes run with by default.
For example, containers running legacy binaries that expect hash
instructions to act as NOPs could configure the reset value of the
container root to control the default reset value for all members of
the container.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
[mpe: Add missing SPDX tag on dexcr.c]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-4-bgray@linux.ibm.com
11 months agopowerpc/dexcr: Track the DEXCR per-process
Benjamin Gray [Wed, 17 Apr 2024 11:23:18 +0000 (21:23 +1000)]
powerpc/dexcr: Track the DEXCR per-process

Add capability to make the DEXCR act as a per-process SPR.

We do not yet have an interface for changing the values per task. We
also expect the kernel to use a single DEXCR value across all tasks
while in privileged state, so there is no need to synchronize after
changing it (the userspace aspects will synchronize upon returning to
userspace).

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-3-bgray@linux.ibm.com
11 months agoselftests/powerpc/dexcr: Add -no-pie to hashchk tests
Benjamin Gray [Wed, 17 Apr 2024 11:23:17 +0000 (21:23 +1000)]
selftests/powerpc/dexcr: Add -no-pie to hashchk tests

The hashchk tests want to verify that the hash key is changed over exec.
It does so by calculating hashes at the same address across an exec.
This is made simpler by disabling PIE functionality, so we can
re-execute ourselves and be using the same addresses in the child.

While -fno-pie is already added, -no-pie is also required.

Fixes: bdb07f35a52f ("selftests/powerpc/dexcr: Add hashst/hashchk test")
Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240417112325.728010-2-bgray@linux.ibm.com
11 months agopowerpc/module: Remove arch specific module bug stuff
Dr. David Alan Gilbert [Fri, 3 May 2024 00:23:17 +0000 (01:23 +0100)]
powerpc/module: Remove arch specific module bug stuff

The last function to reference module_bug_list went in 2008's
  commit b9754568ef17 ("powerpc: Remove dead module_find_bug code")
but I don't think that was called since 2006's
  commit 73c9ceab40b1 ("[POWERPC] Generic BUG for powerpc")

Now that the list has gone, I think we can also clean up the bug
entries in mod_arch_specific.

Lightly boot tested.

Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240503002317.183500-1-linux@treblig.org
11 months agopowerpc/crash: remove unnecessary NULL check before kvfree()
Sourabh Jain [Thu, 2 May 2024 18:20:40 +0000 (23:50 +0530)]
powerpc/crash: remove unnecessary NULL check before kvfree()

Fix the following coccicheck build warning:

arch/powerpc/kexec/crash.c:488:2-8: WARNING: NULL check before some
freeing functions is not needed.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404261048.skfV5DDB-lkp@intel.com/
Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240502182040.774759-1-sourabhjain@linux.ibm.com
11 months agoMAINTAINERS: MMU GATHER: Update Aneesh's address
Michael Ellerman [Tue, 30 Apr 2024 04:43:27 +0000 (14:43 +1000)]
MAINTAINERS: MMU GATHER: Update Aneesh's address

Aneesh's IBM address no longer works, switch to his preferred kernel.org
address.

Acked-by: Aneesh Kumar K.V (Arm) <aneesh.kumar@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240430044327.49363-1-mpe@ellerman.id.au
11 months agoMAINTAINERS: powerpc: Remove Aneesh
Michael Ellerman [Tue, 30 Apr 2024 04:42:28 +0000 (14:42 +1000)]
MAINTAINERS: powerpc: Remove Aneesh

Aneesh is stepping down from powerpc maintenance.

Acked-by: Aneesh Kumar K.V (Arm) <aneesh.kumar@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240430044228.49015-1-mpe@ellerman.id.au
11 months agopowerpc: Mark memory_limit as initdata
Michael Ellerman [Mon, 22 Apr 2024 11:52:31 +0000 (21:52 +1000)]
powerpc: Mark memory_limit as initdata

The `memory_limit` variable should only be used during boot, enforce
that by marking it initdata.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422115231.1769984-1-mpe@ellerman.id.au
11 months agomacintosh/macio-adb: replace of_node_put() with __free
sundar [Wed, 24 Apr 2024 15:07:18 +0000 (20:37 +0530)]
macintosh/macio-adb: replace of_node_put() with __free

use the new cleanup magic to replace of_node_put() with
__free(device_node) marking to auto release when they get out of scope.

Suggested-by: Julia Lawall <julia.lawall@inria.fr>
Signed-off-by: sundar <prosunofficial@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240424150718.5006-1-prosunofficial@gmail.com
11 months agoselftests/powerpc: Install tests in sub-directories
Michael Ellerman [Mon, 22 Apr 2024 13:34:53 +0000 (23:34 +1000)]
selftests/powerpc: Install tests in sub-directories

The sources for the powerpc selftests are arranged into sub-directories.
However when the tests are built and installed, the sub-directories are
squashed, losing the structure.

For example, with the current code the result of installing the selftests is:

  $ tree tools/testing/selftests/kselftest_install
  tools/testing/selftests/kselftest_install
  ├── kselftest
  │   ├── ktap_helpers.sh
  │   ├── module.sh
  │   ├── prefix.pl
  │   └── runner.sh
  ├── kselftest-list.txt
  ├── powerpc
  │   ├── alignment_handler
  │   ├── attr_test
  │   ├── back_to_back_ebbs_test
  │   ├── bad_accesses
  │   ├── bhrb_filter_map_test
  │   ├── bhrb_no_crash_wo_pmu_test
  │   ├── blacklisted_events_test
  │   ├── cache_shape
  │   ├── close_clears_pmcc_test
  │   ├── context_switch
  │   ├── copy_first_unaligned
  ...
  │   ├── settings
  ...
  │   └── wild_bctr
  └── run_kselftest.sh

All the powerpc tests are squashed into the single powerpc directory. In
particular, note that there is a single `settings` file, even though
there are multiple settings files in the powerpc selftest sources. One
of the settings files ends up installed, depending on install order,
even if they have different contents.

Similarly if there were two tests with the same name in different
sub-directories they would clobber each other.

Fix it by replicating the directory structure of the source tree into
the install directory. The result being for example:

  $ tree tools/testing/selftests/kselftest_install
  tools/testing/selftests/kselftest_install
  ├── kselftest
  │   ├── ktap_helpers.sh
  │   ├── module.sh
  │   ├── prefix.pl
  │   └── runner.sh
  ├── kselftest-list.txt
  ├── powerpc
  │   ├── alignment
  │   │   ├── alignment_handler
  │   │   └── copy_first_unaligned
  │   ├── benchmarks
  │   │   ├── context_switch
  │   │   ├── exec_target
  │   │   ├── fork
  │   │   ├── futex_bench
  │   │   ├── gettimeofday
  │   │   ├── mmap_bench
  │   │   ├── null_syscall
  │   │   └── settings
  ...
  │   ├── eeh
  │   │   ├── eeh-basic.sh
  │   │   ├── eeh-functions.sh
  │   │   └── settings
  ...
  │   └── vphn
  │       └── test-vphn
  └── run_kselftest.sh

Note multiple settings files in different sub-directories.

This change also has the effect of changing the names of the tests from
the point of view of the kselftest runner. Before the tests are named
eg:

  powerpc:copy_first_unaligned
  powerpc:cache_shape
  powerpc:reg_access_test

After, the test collection names include the sub-directory:

  powerpc/alignment:copy_first_unaligned
  powerpc/cache_shape:cache_shape
  powerpc/pmu/ebb:reg_access_test

That means whereas previously all powerpc tests could be run with:

  $ ./run_kselftest.sh -c powerpc

After the change it's necessary to pass a regex that matches all powerpc
entries, eg:

  $ ./run_kselftest.sh -c "powerpc.*"

The latter form also works before and after the change.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422133453.1793988-2-mpe@ellerman.id.au
11 months agoselftests/powerpc: Convert pmu Makefile to for loop style
Michael Ellerman [Mon, 22 Apr 2024 13:34:52 +0000 (23:34 +1000)]
selftests/powerpc: Convert pmu Makefile to for loop style

The pmu Makefile has grown more sub directories over the years. Rather
than open coding the rules for each subdir, use for loops.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422133453.1793988-1-mpe@ellerman.id.au
11 months agoselftests/powerpc: make sub-folders buildable on their own
Madhavan Srinivasan [Thu, 29 Feb 2024 09:37:11 +0000 (15:07 +0530)]
selftests/powerpc: make sub-folders buildable on their own

Build breaks when executing make with run_tests for sub-folders
under powerpc. This is because, CFLAGS and GIT_VERSION macros are
defined in Makefile of toplevel powerpc folder.

  make: Entering directory '/home/maddy/linux/tools/testing/selftests/powerpc/mm'
  gcc     hugetlb_vs_thp_test.c ../harness.c ../utils.c  -o /home/maddy/selftest_output//hugetlb_vs_thp_test
  hugetlb_vs_thp_test.c:6:10: fatal error: utils.h: No such file or directory
      6 | #include "utils.h"
        |          ^~~~~~~~~
  compilation terminated.

Fix this by adding the flags.mk in each sub-folder Makefile. Also remove
the CFLAGS and GIT_VERSION macros from powerpc/ folder Makefile since
the same is definied in flags.mk

Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240229093711.581230-3-maddy@linux.ibm.com
11 months agoselftests/powerpc: Add flags.mk to support pmu buildable
Madhavan Srinivasan [Thu, 29 Feb 2024 09:37:10 +0000 (15:07 +0530)]
selftests/powerpc: Add flags.mk to support pmu buildable

When running `make -C powerpc/pmu run_tests` from top level selftests
directory, currently this error is being reported:

  make: Entering directory '/home/maddy/linux/tools/testing/selftests/powerpc/pmu'
  Makefile:40: warning: overriding recipe for target 'emit_tests'
  ../../lib.mk:111: warning: ignoring old recipe for target 'emit_tests'
  gcc -m64    count_instructions.c ../harness.c event.c lib.c ../utils.c loop.S  -o /home/maddy/selftest_output//count_instructions
  In file included from count_instructions.c:13:
  event.h:12:10: fatal error: utils.h: No such file or directory
  12 | #include "utils.h"
    |          ^~~~~~~~~
  compilation terminated.

This is due to missing of include path in CFLAGS. That is, CFLAGS and
GIT_VERSION macros are defined in the powerpc/ folder Makefile which
in this case is not involved.

To address the failure in case of executing specific sub-folder test
directly, a new rule file has been addded by the patch called "flags.mk"
under selftest/powerpc/ folder and is linked to all the Makefile of
powerpc/pmu sub-folders.

Reported-by: Sachin Sant <sachinp@linux.ibm.com>
Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Tested-by: Sachin Sant <sachinp@linux.ibm.com>
[mpe: Fixup ifeq, make GIT_VERSION simply expanded to avoid re-executing git describe]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240229093711.581230-2-maddy@linux.ibm.com
11 months agoselftests/powerpc: Re-order *FLAGS to follow lib.mk
Madhavan Srinivasan [Thu, 29 Feb 2024 09:37:09 +0000 (15:07 +0530)]
selftests/powerpc: Re-order *FLAGS to follow lib.mk

In some powerpc/ sub-folder Makefiles, CFLAGS are defined before lib.mk
include. Clean it up by re-ordering the flags to follow after the mk
include. This is needed to support sub-folders in powerpc/ buildable on
its own.

Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240229093711.581230-1-maddy@linux.ibm.com
11 months agopowerpc/pseries/vio: Don't return ENODEV if node or compatible missing
Lidong Zhong [Thu, 11 Apr 2024 02:04:50 +0000 (10:04 +0800)]
powerpc/pseries/vio: Don't return ENODEV if node or compatible missing

We noticed the following nuisance messages during boot process:

  vio vio: uevent: failed to send synthetic uevent
  vio 4000: uevent: failed to send synthetic uevent
  vio 4001: uevent: failed to send synthetic uevent
  vio 4002: uevent: failedto send synthetic uevent
  vio 4004: uevent: failed to send synthetic uevent

It's caused by either vio_register_device_node() failing to set
dev->of_node or the node is missing a "compatible" property. To match
the definition of modalias in modalias_show(), remove the return of
ENODEV in such cases. The failure messages is also suppressed with this
change.

Signed-off-by: Lidong Zhong <lidong.zhong@suse.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240411020450.12725-1-lidong.zhong@suse.com
11 months agopowerpc/pseries: Enforce hcall result buffer validity and size
Nathan Lynch [Mon, 8 Apr 2024 14:08:31 +0000 (09:08 -0500)]
powerpc/pseries: Enforce hcall result buffer validity and size

plpar_hcall(), plpar_hcall9(), and related functions expect callers to
provide valid result buffers of certain minimum size. Currently this
is communicated only through comments in the code and the compiler has
no idea.

For example, if I write a bug like this:

  long retbuf[PLPAR_HCALL_BUFSIZE]; // should be PLPAR_HCALL9_BUFSIZE
  plpar_hcall9(H_ALLOCATE_VAS_WINDOW, retbuf, ...);

This compiles with no diagnostics emitted, but likely results in stack
corruption at runtime when plpar_hcall9() stores results past the end
of the array. (To be clear this is a contrived example and I have not
found a real instance yet.)

To make this class of error less likely, we can use explicitly-sized
array parameters instead of pointers in the declarations for the hcall
APIs. When compiled with -Warray-bounds[1], the code above now
provokes a diagnostic like this:

error: array argument is too small;
is of size 32, callee requires at least 72 [-Werror,-Warray-bounds]
   60 |                 plpar_hcall9(H_ALLOCATE_VAS_WINDOW, retbuf,
      |                 ^                                   ~~~~~~

[1] Enabled for LLVM builds but not GCC for now. See commit
    0da6e5fd6c37 ("gcc: disable '-Warray-bounds' for gcc-13 too") and
    related changes.

Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240408-pseries-hvcall-retbuf-v1-1-ebc73d7253cf@linux.ibm.com
11 months agopowerpc/dart: Drop unnecessary call to kmemleak_no_scan()
Michael Ellerman [Fri, 19 Apr 2024 11:59:13 +0000 (21:59 +1000)]
powerpc/dart: Drop unnecessary call to kmemleak_no_scan()

Erhard reported that kmemleak was showing a warning at boot:

  kmemleak: Not scanning unknown object at 0xc00000007f000000
  CPU: 0 PID: 0 Comm: swapper Not tainted 5.19.0-rc3-PMacG5+ #2
  Call Trace:
   .dump_stack_lvl+0x7c/0xc4 (unreliable)
   .kmemleak_no_scan+0xe0/0x100
   .iommu_init_early_dart+0x2f0/0x924
   .pmac_probe+0x1b0/0x20c
   .setup_arch+0x1b8/0x674
   .start_kernel+0xdc/0xb74
   start_here_common+0x1c/0x44
  DART table allocated at: (____ptrval____)

Which he bisected to a change in kmemleak, commit
23c2d497de21 ("mm: kmemleak: take a full lowmem check in kmemleak_*_phys()").

Because pmac_probe() is called before mem_topology_setup(), the min/
max PFN variables are still zero. That causes kmemleak_alloc_phys() to
ignore the allocation, because the checks against the PFN fail. Then
kmemleak_no_scan() can't find the allocation and prints warning.

Given that kmemleak_alloc_phys() is ignoring the allocation to begin
with, there's no need to call kmemleak_no_scan() at all, which avoids
the warning.

Reported-by: Erhard Furtner <erhard_f@mailbox.org>
Closes: https://lore.kernel.org/all/bug-216156-206035@https.bugzilla.kernel.org%2F/
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240419115913.3317575-1-mpe@ellerman.id.au
11 months agopowerpc/eeh: Permanently disable the removed device
Ganesh Goudar [Mon, 22 Apr 2024 07:57:37 +0000 (13:27 +0530)]
powerpc/eeh: Permanently disable the removed device

When a device is hot removed on powernv, the hotplug driver clears
the device's state. However, on pseries, if a device is removed by
phyp after reaching the error threshold, the kernel remains unaware,
leading to the device not being torn down. This prevents necessary
remediation actions like failover.

Permanently disable the device if the presence check fails.

Also, in eeh_dev_check_failure in we may consider the error as false
positive if the device is hotpluged out as the get_state call returns
EEH_STATE_NOT_SUPPORT and we may end up not clearing the device state,
so log the event if the state is not moved to permanent failure state.

Signed-off-by: Ganesh Goudar <ganeshgr@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422075737.1405551-1-ganeshgr@linux.ibm.com
11 months agoDocumentation/powerpc: update fadump implementation details
Sourabh Jain [Mon, 22 Apr 2024 19:59:32 +0000 (01:29 +0530)]
Documentation/powerpc: update fadump implementation details

The patch titled ("powerpc: make fadump resilient with memory add/remove
events") has made significant changes to the implementation of fadump,
particularly on elfcorehdr creation and fadump crash info header
structure. Therefore, updating the fadump implementation documentation
to reflect those changes.

Following updates are done to firmware assisted dump documentation:

1. The elfcorehdr is no longer stored after fadump HDR in the reserved
   dump area. Instead, the second kernel dynamically allocates memory
   for the elfcorehdr within the address range from 0 to the boot memory
   size. Therefore, update figures 1 and 2 of Memory Reservation during
   the first and second kernels to reflect this change.

2. A version field has been added to the fadump header to manage the
   future changes to fadump crash info header structure without changing
   the fadump header magic number in the future. Therefore, remove the
   corresponding TODO from the document.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422195932.1583833-4-sourabhjain@linux.ibm.com
11 months agopowerpc/fadump: add hotplug_ready sysfs interface
Sourabh Jain [Mon, 22 Apr 2024 19:59:31 +0000 (01:29 +0530)]
powerpc/fadump: add hotplug_ready sysfs interface

The elfcorehdr describes the CPUs and memory of the crashed kernel to
the kernel that captures the dump, known as the second or fadump kernel.
The elfcorehdr needs to be updated if the system's memory changes due to
memory hotplug or online/offline events.

Currently, memory hotplug events are monitored in userspace by udev
rules, and fadump is re-registered, which recreates the elfcorehdr with
the latest available memory in the system.

However, the previous patch ("powerpc: make fadump resilient with memory
add/remove events") moved the creation of elfcorehdr to the second or
fadump kernel. This eliminates the need to regenerate the elfcorehdr
during memory hotplug or online/offline events.

Create a sysfs entry at /sys/kernel/fadump/hotplug_ready to let
userspace know that fadump re-registration is not required for memory
add/remove events.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422195932.1583833-3-sourabhjain@linux.ibm.com
11 months agopowerpc: make fadump resilient with memory add/remove events
Sourabh Jain [Mon, 22 Apr 2024 19:59:30 +0000 (01:29 +0530)]
powerpc: make fadump resilient with memory add/remove events

Due to changes in memory resources caused by either memory hotplug or
online/offline events, the elfcorehdr, which describes the CPUs and
memory of the crashed kernel to the kernel that collects the dump (known
as second/fadump kernel), becomes outdated. Consequently, attempting
dump collection with an outdated elfcorehdr can lead to failed or
inaccurate dump collection.

Memory hotplug or online/offline events is referred as memory add/remove
events in reset of the commit message.

The current solution to address the aforementioned issue is as follows:
Monitor memory add/remove events in userspace using udev rules, and
re-register fadump whenever there are changes in memory resources. This
leads to the creation of a new elfcorehdr with updated system memory
information.

There are several notable issues associated with re-registering fadump
for every memory add/remove events.

1. Bulk memory add/remove events with udev-based fadump re-registration
   can lead to race conditions and, more importantly, it creates a wide
   window during which fadump is inactive until all memory add/remove
   events are settled.
2. Re-registering fadump for every memory add/remove event is
   inefficient.
3. The memory for elfcorehdr is allocated based on the memblock regions
   available during early boot and remains fixed thereafter. However, if
   elfcorehdr is later recreated with additional memblock regions, its
   size will increase, potentially leading to memory corruption.

Address the aforementioned challenges by shifting the creation of
elfcorehdr from the first kernel (also referred as the crashed kernel),
where it was created and frequently recreated for every memory
add/remove event, to the fadump kernel. As a result, the elfcorehdr only
needs to be created once, thus eliminating the necessity to re-register
fadump during memory add/remove events.

At present, the first kernel prepares fadump header and stores it in the
fadump reserved area. The fadump header includes the start address of
the elfcorehdr, crashing CPU details, and other relevant information. In
the event of a crash in the first kernel, the second/fadump boots and
accesses the fadump header prepared by the first kernel. It then
performs the following steps in a platform-specific function
[rtas|opal]_fadump_process:

1. Sanity check for fadump header
2. Update CPU notes in elfcorehdr

Along with the above, update the setup_fadump()/fadump.c to create
elfcorehdr and set its address to the global variable elfcorehdr_addr
for the vmcore module to process it in the second/fadump kernel.

Section below outlines the information required to create the elfcorehdr
and the changes made to make it available to the fadump kernel if it's
not already.

To create elfcorehdr, the following crashed kernel information is
required: CPU notes, vmcoreinfo, and memory ranges.

At present, the CPU notes are already prepared in the fadump kernel, so
no changes are needed in that regard. The fadump kernel has access to
all crashed kernel memory regions, including boot memory regions that
are relocated by firmware to fadump reserved areas, so no changes for
that either. However, it is necessary to add new members to the fadump
header, i.e., the 'fadump_crash_info_header' structure, in order to pass
the crashed kernel's vmcoreinfo address and its size to fadump kernel.

In addition to the vmcoreinfo address and size, there are a few other
attributes also added to the fadump_crash_info_header structure.

1. version:
   It stores the fadump header version, which is currently set to 1.
   This provides flexibility to update the fadump crash info header in
   the future without changing the magic number. For each change in the
   fadump header, the version will be increased. This will help the
   updated kernel determine how to handle kernel dumps from older
   kernels. The magic number remains relevant for checking fadump header
   corruption.

2. pt_regs_sz/cpu_mask_sz:
   Store size of pt_regs and cpu_mask structure of first kernel. These
   attributes are used to prevent dump processing if the sizes of
   pt_regs or cpu_mask structure differ between the first and fadump
   kernels.

Note: if either first/crashed kernel or second/fadump kernel do not have
the changes introduced here then kernel fail to collect the dump and
prints relevant error message on the console.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240422195932.1583833-2-sourabhjain@linux.ibm.com
11 months agopowerpc/pseries: Add failure related checks for h_get_mpp and h_get_ppp
Shrikanth Hegde [Fri, 12 Apr 2024 09:20:47 +0000 (14:50 +0530)]
powerpc/pseries: Add failure related checks for h_get_mpp and h_get_ppp

Couple of Minor fixes:

- hcall return values are long. Fix that for h_get_mpp, h_get_ppp and
parse_ppp_data

- If hcall fails, values set should be at-least zero. It shouldn't be
uninitialized values. Fix that for h_get_mpp and h_get_ppp

Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240412092047.455483-3-sshegde@linux.ibm.com
11 months agopowerpc/pseries: Add pool idle time at LPAR boot
Shrikanth Hegde [Fri, 12 Apr 2024 09:20:46 +0000 (14:50 +0530)]
powerpc/pseries: Add pool idle time at LPAR boot

When there are no options specified for lparstat, it is expected to
give reports since LPAR(Logical Partition) boot.

APP(Available Processor Pool) is an indicator of how many cores in the
shared pool are free to use in Shared Processor LPAR(SPLPAR). APP is
derived using pool_idle_time which is obtained using H_PIC call.

The interval based reports show correct APP value while since boot
report shows very high APP values. This happens because in that case APP
is obtained by dividing pool idle time by LPAR uptime. Since pool idle
time is reported by the PowerVM hypervisor since its boot, it need not
align with LPAR boot.

To fix that export boot pool idle time in lparcfg and powerpc-utils will
use this info to derive APP as below for since boot reports.

APP = (pool idle time - boot pool idle time) / (uptime * timebase)

Results:: Observe APP values.
====================== Shared LPAR ================================
lparstat
System Configuration
type=Shared mode=Uncapped smt=8 lcpu=12 mem=15573440 kB cpus=37 ent=12.00

reboot
stress-ng --cpu=$(nproc) -t 600
sleep 600
So in this case app is expected to close to 37-6=31.

====== 6.9-rc1 and lparstat 1.3.10  =============
%user  %sys %wait    %idle    physc %entc lbusy   app  vcsw phint
----- ----- -----    -----    ----- ----- ----- ----- ----- -----
47.48  0.01  0.00    52.51     0.00  0.00 47.49 69099.72 541547    21

=== With this patch and powerpc-utils patch to do the above equation ===
%user  %sys %wait    %idle    physc %entc lbusy   app  vcsw phint
----- ----- -----    -----    ----- ----- ----- ----- ----- -----
47.48  0.01  0.00    52.51     5.73 47.75 47.49 31.21 541753    21
=====================================================================

Note: physc, purr/idle purr being inaccurate is being handled in a
separate patch in powerpc-utils tree.

Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240412092047.455483-2-sshegde@linux.ibm.com
11 months agopowerpc/crash: add crash memory hotplug support
Sourabh Jain [Tue, 26 Mar 2024 05:54:13 +0000 (11:24 +0530)]
powerpc/crash: add crash memory hotplug support

Extend the arch crash hotplug handler, as introduced by the patch title
("powerpc: add crash CPU hotplug support"), to also support memory
add/remove events.

Elfcorehdr describes the memory of the crash kernel to capture the
kernel; hence, it needs to be updated if memory resources change due to
memory add/remove events. Therefore, arch_crash_handle_hotplug_event()
is updated to recreate the elfcorehdr and replace it with the previous
one on memory add/remove events.

The memblock list is used to prepare the elfcorehdr. In the case of
memory hot remove, the memblock list is updated after the arch crash
hotplug handler is triggered, as depicted in Figure 1. Thus, the
hot-removed memory is explicitly removed from the crash memory ranges
to ensure that the memory ranges added to elfcorehdr do not include the
hot-removed memory.

    Memory remove
          |
          v
    Offline pages
          |
          v
 Initiate memory notify call <----> crash hotplug handler
 chain for MEM_OFFLINE event
          |
          v
 Update memblock list

  Figure 1

There are two system calls, `kexec_file_load` and `kexec_load`, used to
load the kdump image. A few changes have been made to ensure that the
kernel can safely update the elfcorehdr component of the kdump image for
both system calls.

For the kexec_file_load syscall, kdump image is prepared in the kernel.
To support an increasing number of memory regions, the elfcorehdr is
built with extra buffer space to ensure that it can accommodate
additional memory ranges in future.

For the kexec_load syscall, the elfcorehdr is updated only if the
KEXEC_CRASH_HOTPLUG_SUPPORT kexec flag is passed to the kernel by the
kexec tool. Passing this flag to the kernel indicates that the
elfcorehdr is built to accommodate additional memory ranges and the
elfcorehdr segment is not considered for SHA calculation, making it safe
to update.

The changes related to this feature are kept under the CRASH_HOTPLUG
config, and it is enabled by default.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-7-sourabhjain@linux.ibm.com
11 months agopowerpc/crash: add crash CPU hotplug support
Sourabh Jain [Tue, 26 Mar 2024 05:54:12 +0000 (11:24 +0530)]
powerpc/crash: add crash CPU hotplug support

Due to CPU/Memory hotplug or online/offline events, the elfcorehdr
(which describes the CPUs and memory of the crashed kernel) and FDT
(Flattened Device Tree) of kdump image becomes outdated. Consequently,
attempting dump collection with an outdated elfcorehdr or FDT can lead
to failed or inaccurate dump collection.

Going forward, CPU hotplug or online/offline events are referred as
CPU/Memory add/remove events.

The current solution to address the above issue involves monitoring the
CPU/Memory add/remove events in userspace using udev rules and whenever
there are changes in CPU and memory resources, the entire kdump image
is loaded again. The kdump image includes kernel, initrd, elfcorehdr,
FDT, purgatory. Given that only elfcorehdr and FDT get outdated due to
CPU/Memory add/remove events, reloading the entire kdump image is
inefficient. More importantly, kdump remains inactive for a substantial
amount of time until the kdump reload completes.

To address the aforementioned issue, commit 247262756121 ("crash: add
generic infrastructure for crash hotplug support") added a generic
infrastructure that allows architectures to selectively update the kdump
image component during CPU or memory add/remove events within the kernel
itself.

In the event of a CPU or memory add/remove events, the generic crash
hotplug event handler, `crash_handle_hotplug_event()`, is triggered. It
then acquires the necessary locks to update the kdump image and invokes
the architecture-specific crash hotplug handler,
`arch_crash_handle_hotplug_event()`, to update the required kdump image
components.

This patch adds crash hotplug handler for PowerPC and enable support to
update the kdump image on CPU add/remove events. Support for memory
add/remove events is added in a subsequent patch with the title
"powerpc: add crash memory hotplug support"

As mentioned earlier, only the elfcorehdr and FDT kdump image components
need to be updated in the event of CPU or memory add/remove events.
However, on PowerPC architecture crash hotplug handler only updates the
FDT to enable crash hotplug support for CPU add/remove events. Here's
why.

The elfcorehdr on PowerPC is built with possible CPUs, and thus, it does
not need an update on CPU add/remove events. On the other hand, the FDT
needs to be updated on CPU add events to include the newly added CPU. If
the FDT is not updated and the kernel crashes on a newly added CPU, the
kdump kernel will fail to boot due to the unavailability of the crashing
CPU in the FDT. During the early boot, it is expected that the boot CPU
must be a part of the FDT; otherwise, the kernel will raise a BUG and
fail to boot. For more information, refer to commit 36ae37e3436b0
("powerpc: Make boot_cpuid common between 32 and 64-bit"). Since it is
okay to have an offline CPU in the kdump FDT, no action is taken in case
of CPU removal.

There are two system calls, `kexec_file_load` and `kexec_load`, used to
load the kdump image. Few changes have been made to ensure kernel can
safely update the FDT of kdump image loaded using both system calls.

For kexec_file_load syscall the kdump image is prepared in kernel. So to
support an increasing number of CPUs, the FDT is constructed with extra
buffer space to ensure it can accommodate a possible number of CPU
nodes. Additionally, a call to fdt_pack (which trims the unused space
once the FDT is prepared) is avoided if this feature is enabled.

For the kexec_load syscall, the FDT is updated only if the
KEXEC_CRASH_HOTPLUG_SUPPORT kexec flag is passed to the kernel by
userspace (kexec tools). When userspace passes this flag to the kernel,
it indicates that the FDT is built to accommodate possible CPUs, and the
FDT segment is excluded from SHA calculation, making it safe to update.

The changes related to this feature are kept under the CRASH_HOTPLUG
config, and it is enabled by default.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-6-sourabhjain@linux.ibm.com
11 months agopowerpc/kexec: make the update_cpus_node() function public
Sourabh Jain [Tue, 26 Mar 2024 05:54:11 +0000 (11:24 +0530)]
powerpc/kexec: make the update_cpus_node() function public

Move the update_cpus_node() from kexec/{file_load_64.c => core_64.c}
to allow other kexec components to use it.

Later in the series, this function is used for in-kernel updates
to the kdump image during CPU/memory hotplug or online/offline events for
both kexec_load and kexec_file_load syscalls.

No functional changes are intended.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-5-sourabhjain@linux.ibm.com
11 months agopowerpc/kexec: move *_memory_ranges functions to ranges.c
Sourabh Jain [Tue, 26 Mar 2024 05:54:10 +0000 (11:24 +0530)]
powerpc/kexec: move *_memory_ranges functions to ranges.c

Move the following functions form kexec/{file_load_64.c => ranges.c} and
make them public so that components other than KEXEC_FILE can also use
these functions.
1. get_exclude_memory_ranges
2. get_reserved_memory_ranges
3. get_crash_memory_ranges
4. get_usable_memory_ranges

Later in the series get_crash_memory_ranges function is utilized for
in-kernel updates to kdump image during CPU/Memory hotplug or
online/offline events for both kexec_load and kexec_file_load syscalls.

Since the above functions are moved to ranges.c, some of the helper
functions in ranges.c are no longer required to be public. Mark them as
static and removed them from kexec_ranges.h header file.

Finally, remove the CONFIG_KEXEC_FILE build dependency for range.c
because it is required for other config, such as CONFIG_CRASH_DUMP.

No functional changes are intended.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-4-sourabhjain@linux.ibm.com
11 months agocrash: add a new kexec flag for hotplug support
Sourabh Jain [Tue, 26 Mar 2024 05:54:09 +0000 (11:24 +0530)]
crash: add a new kexec flag for hotplug support

Commit a72bbec70da2 ("crash: hotplug support for kexec_load()")
introduced a new kexec flag, `KEXEC_UPDATE_ELFCOREHDR`. Kexec tool uses
this flag to indicate to the kernel that it is safe to modify the
elfcorehdr of the kdump image loaded using the kexec_load system call.

However, it is possible that architectures may need to update kexec
segments other then elfcorehdr. For example, FDT (Flatten Device Tree)
on PowerPC. Introducing a new kexec flag for every new kexec segment
may not be a good solution. Hence, a generic kexec flag bit,
`KEXEC_CRASH_HOTPLUG_SUPPORT`, is introduced to share the CPU/Memory
hotplug support intent between the kexec tool and the kernel for the
kexec_load system call.

Now we have two kexec flags that enables crash hotplug support for
kexec_load system call. First is KEXEC_UPDATE_ELFCOREHDR (only used in
x86), and second is KEXEC_CRASH_HOTPLUG_SUPPORT (for all architectures).

To simplify the process of finding and reporting the crash hotplug
support the following changes are introduced.

1. Define arch specific function to process the kexec flags and
   determine crash hotplug support

2. Rename the @update_elfcorehdr member of struct kimage to
   @hotplug_support and populate it for both kexec_load and
   kexec_file_load syscalls, because architecture can update more than
   one kexec segment

3. Let generic function crash_check_hotplug_support report hotplug
   support for loaded kdump image based on value of @hotplug_support

To bring the x86 crash hotplug support in line with the above points,
the following changes have been made:

- Introduce the arch_crash_hotplug_support function to process kexec
  flags and determine crash hotplug support

- Remove the arch_crash_hotplug_[cpu|memory]_support functions

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Baoquan He <bhe@redhat.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-3-sourabhjain@linux.ibm.com
11 months agocrash: forward memory_notify arg to arch crash hotplug handler
Sourabh Jain [Tue, 26 Mar 2024 05:54:08 +0000 (11:24 +0530)]
crash: forward memory_notify arg to arch crash hotplug handler

In the event of memory hotplug or online/offline events, the crash
memory hotplug notifier `crash_memhp_notifier()` receives a
`memory_notify` object but doesn't forward that object to the
generic and architecture-specific crash hotplug handler.

The `memory_notify` object contains the starting PFN (Page Frame Number)
and the number of pages in the hot-removed memory. This information is
necessary for architectures like PowerPC to update/recreate the kdump
image, specifically `elfcorehdr`.

So update the function signature of `crash_handle_hotplug_event()` and
`arch_crash_handle_hotplug_event()` to accept the `memory_notify` object
as an argument from crash memory hotplug notifier.

Since no such object is available in the case of CPU hotplug event, the
crash CPU hotplug notifier `crash_cpuhp_online()` passes NULL to the
crash hotplug handler.

Signed-off-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Acked-by: Baoquan He <bhe@redhat.com>
Acked-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326055413.186534-2-sourabhjain@linux.ibm.com
11 months agopowerpc/mm: Update the memory limit based on direct mapping restrictions
Aneesh Kumar K.V (IBM) [Wed, 3 Apr 2024 08:36:11 +0000 (14:06 +0530)]
powerpc/mm: Update the memory limit based on direct mapping restrictions

memory limit value specified by the user are further updated such that
the value is 16MB aligned. This is because hash translation mode use
16MB as direct mapping page size. Make sure we update the global
variable 'memory_limit' with the 16MB aligned value such that all kernel
components will see the new aligned value of the memory limit.

Signed-off-by: Aneesh Kumar K.V (IBM) <aneesh.kumar@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240403083611.172833-3-aneesh.kumar@kernel.org
11 months agopowerpc/fadump: Don't update the user-specified memory limit
Aneesh Kumar K.V (IBM) [Wed, 3 Apr 2024 08:36:10 +0000 (14:06 +0530)]
powerpc/fadump: Don't update the user-specified memory limit

If the user specifies the memory limit, the kernel should honor it such
that all allocation and reservations are made within the memory limit
specified. fadump was breaking that rule. Remove the code which updates
the memory limit such that fadump reservations are done within the
limit specified.

Signed-off-by: Aneesh Kumar K.V (IBM) <aneesh.kumar@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240403083611.172833-2-aneesh.kumar@kernel.org
11 months agopowerpc/mm: Align memory_limit value specified using mem= kernel parameter
Aneesh Kumar K.V (IBM) [Wed, 3 Apr 2024 08:36:09 +0000 (14:06 +0530)]
powerpc/mm: Align memory_limit value specified using mem= kernel parameter

The value specified for the memory limit is used to set a restriction on
memory usage. It is important to ensure that this restriction is within
the linear map kernel address space range. The hash page table
translation uses a 16MB page size to map the kernel linear map address
space. htab_bolt_mapping() function aligns down the size of the range
while mapping kernel linear address space. Since the memblock limit is
enforced very early during boot, before we can detect the type of memory
translation (radix vs hash), we align the memory limit value specified
as a kernel parameter to 16MB. This alignment value will work for both
hash and radix translations.

Signed-off-by: Aneesh Kumar K.V (IBM) <aneesh.kumar@kernel.org>
Acked-by: Joel Savitz <jsavitz@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240403083611.172833-1-aneesh.kumar@kernel.org
11 months agopowerpc/ptdump: Fix walk_vmemmap() to also print first vmemmap entry
Ritesh Harjani (IBM) [Wed, 17 Apr 2024 15:07:40 +0000 (20:37 +0530)]
powerpc/ptdump: Fix walk_vmemmap() to also print first vmemmap entry

Currently walk_vmemmap() skips the first vmemmap entry pointed to by
vmemmap_list pointer itself. Fix that.

With the fix applied the vmemmap entry at 0xc00c000000000000 for hash is
displayed:

  $ cat /sys/kernel/debug/kernel_hash_pagetable
  ...
  0xc00c000000010000:     AVPN:cd7bd4e0000          ssize: 1T ...
  0xc00c000000000000:     AVPN:cd7bd4e0000          ssize: 1T ...

Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
[mpe: Tweak change log wording and add example output]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/a19ee3dc2b304d39da364a592d5cd167449f8c4a.1713365940.git.ritesh.list@gmail.com
12 months agopowerpc: Avoid nmi_enter/nmi_exit in real mode interrupt.
Mahesh Salgaonkar [Wed, 10 Apr 2024 04:30:06 +0000 (10:00 +0530)]
powerpc: Avoid nmi_enter/nmi_exit in real mode interrupt.

nmi_enter()/nmi_exit() touches per cpu variables which can lead to kernel
crash when invoked during real mode interrupt handling (e.g. early HMI/MCE
interrupt handler) if percpu allocation comes from vmalloc area.

Early HMI/MCE handlers are called through DEFINE_INTERRUPT_HANDLER_NMI()
wrapper which invokes nmi_enter/nmi_exit calls. We don't see any issue when
percpu allocation is from the embedded first chunk. However with
CONFIG_NEED_PER_CPU_PAGE_FIRST_CHUNK enabled there are chances where percpu
allocation can come from the vmalloc area.

With kernel command line "percpu_alloc=page" we can force percpu allocation
to come from vmalloc area and can see kernel crash in machine_check_early:

[    1.215714] NIP [c000000000e49eb4] rcu_nmi_enter+0x24/0x110
[    1.215717] LR [c0000000000461a0] machine_check_early+0xf0/0x2c0
[    1.215719] --- interrupt: 200
[    1.215720] [c000000fffd73180] [0000000000000000] 0x0 (unreliable)
[    1.215722] [c000000fffd731b0] [0000000000000000] 0x0
[    1.215724] [c000000fffd73210] [c000000000008364] machine_check_early_common+0x134/0x1f8

Fix this by avoiding use of nmi_enter()/nmi_exit() in real mode if percpu
first chunk is not embedded.

Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Tested-by: Shirisha Ganta <shirisha@linux.ibm.com>
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240410043006.81577-1-mahesh@linux.ibm.com
12 months agopowerpc: Add static_key_feature_checks_initialized flag
Nicholas Miehlbradt [Mon, 8 Apr 2024 05:23:58 +0000 (05:23 +0000)]
powerpc: Add static_key_feature_checks_initialized flag

JUMP_LABEL_FEATURE_CHECK_DEBUG used static_key_intialized to determine
whether {cpu,mmu}_has_feature() is used before static keys were
initialized. However, {cpu,mmu}_has_feature() should not be used before
setup_feature_keys() is called but static_key_initialized is set well
before this by the call to jump_label_init() in early_init_devtree().
This creates a window in which JUMP_LABEL_FEATURE_CHECK_DEBUG will not
detect misuse and report errors. Add a flag specifically to indicate
when {cpu,mmu}_has_feature() is safe to use.

Signed-off-by: Nicholas Miehlbradt <nicholas@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240408052358.5030-1-nicholas@linux.ibm.com
12 months agopowerpc/52xx: Replace of_gpio.h by proper one
Andy Shevchenko [Wed, 13 Mar 2024 13:56:45 +0000 (15:56 +0200)]
powerpc/52xx: Replace of_gpio.h by proper one

of_gpio.h is deprecated and subject to remove.
The driver doesn't use it directly, replace it
with what is really being used.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240313135645.2066362-1-andriy.shevchenko@linux.intel.com
12 months agopowerpc: Fix fatal warnings flag for LLVM's integrated assembler
Nathan Chancellor [Fri, 5 Apr 2024 19:31:22 +0000 (12:31 -0700)]
powerpc: Fix fatal warnings flag for LLVM's integrated assembler

When building with LLVM_IAS=1, there is an error because
'-fatal-warnings' is not recognized as a valid flag:

  clang: error: unsupported argument '-fatal-warnings' to option '-Wa,'

Use the double hyphen version of the flag, '--fatal-warnings', which
works with both the GNU assembler and LLVM's integrated assembler.

Fixes: 608d4a5ca563 ("powerpc: Error on assembly warnings")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Justin Stitt <justinstitt@google.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240405-ppc-fix-wa-fatal-warnings-clang-v1-1-bdcd969f2ef0@kernel.org
12 months agopowerpc: Fix PS3 allmodconfig warning
Geoff Levand [Mon, 1 Apr 2024 07:08:31 +0000 (16:08 +0900)]
powerpc: Fix PS3 allmodconfig warning

The struct ps3_notification_device in the ps3_probe_thread routine
is too large to be on the stack, causing a warning for an
allmodconfig build with clang.

Change the struct ps3_notification_device from a variable on the stack
to a dynamically allocated variable.

Reported-by: Arnd Bergmann <arnd@kernel.org>
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/d64f06f4-81ae-4ec5-ab3b-d7f7f091e0ac@infradead.org
12 months agopowerpc: Error on assembly warnings
Benjamin Gray [Tue, 26 Mar 2024 04:44:20 +0000 (15:44 +1100)]
powerpc: Error on assembly warnings

We currently enable -Werror on the arch/powerpc subtree. However this
only catches C warnings. Assembly warnings are logged, but the make
invocation will still succeed. This can allow incorrect syntax such as

  ori r3, r4, r5

to be compiled without catching that the assembler is treating r5
as the immediate value 5.

To prevent this in assembly files and inline assembly, add the
-fatal-warnings option to assembler invocations.

Signed-off-by: Benjamin Gray <bgray@linux.ibm.com>
Tested-by: Andrew Donnellan <ajd@linux.ibm.com>
Reviewed-by: Andrew Donnellan <ajd@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240326044420.577031-1-bgray@linux.ibm.com
12 months agopowerpc/fsl-soc: hide unused const variable
Arnd Bergmann [Wed, 3 Apr 2024 08:06:19 +0000 (10:06 +0200)]
powerpc/fsl-soc: hide unused const variable

vmpic_msi_feature is only used conditionally, which triggers a rare
-Werror=unused-const-variable= warning with gcc:

arch/powerpc/sysdev/fsl_msi.c:567:37: error: 'vmpic_msi_feature' defined but not used [-Werror=unused-const-variable=]
  567 | static const struct fsl_msi_feature vmpic_msi_feature =

Hide this one in the same #ifdef as the reference so we can turn on
the warning by default.

Fixes: 305bcf26128e ("powerpc/fsl-soc: use CONFIG_EPAPR_PARAVIRT for hcalls")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240403080702.3509288-2-arnd@kernel.org
12 months agopowerpc: Use str_plural() in cpu_init_thread_core_maps()
Thorsten Blum [Sun, 31 Mar 2024 22:22:50 +0000 (00:22 +0200)]
powerpc: Use str_plural() in cpu_init_thread_core_maps()

Fixes the following Coccinelle/coccicheck warning reported by
string_choices.cocci:

opportunity for str_plural(tpc)

Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240331222249.107467-2-thorsten.blum@toblux.com
12 months agoLinux 6.9-rc2
Linus Torvalds [Sun, 31 Mar 2024 21:32:39 +0000 (14:32 -0700)]
Linux 6.9-rc2

12 months agoMerge tag 'kbuild-fixes-v6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/masahi...
Linus Torvalds [Sun, 31 Mar 2024 18:23:51 +0000 (11:23 -0700)]
Merge tag 'kbuild-fixes-v6.9' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull Kbuild fixes from Masahiro Yamada:

 - Deduplicate Kconfig entries for CONFIG_CXL_PMU

 - Fix unselectable choice entry in MIPS Kconfig, and forbid this
   structure

 - Remove unused include/asm-generic/export.h

 - Fix a NULL pointer dereference bug in modpost

 - Enable -Woverride-init warning consistently with W=1

 - Drop KCSAN flags from *.mod.c files

* tag 'kbuild-fixes-v6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  kconfig: Fix typo HEIGTH to HEIGHT
  Documentation/llvm: Note s390 LLVM=1 support with LLVM 18.1.0 and newer
  kbuild: Disable KCSAN for autogenerated *.mod.c intermediaries
  kbuild: make -Woverride-init warnings more consistent
  modpost: do not make find_tosym() return NULL
  export.h: remove include/asm-generic/export.h
  kconfig: do not reparent the menu inside a choice block
  MIPS: move unselectable FIT_IMAGE_FDT_EPM5 out of the "System type" choice
  cxl: remove CONFIG_CXL_PMU entry in drivers/cxl/Kconfig

12 months agoMerge tag 'edac_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 18:15:32 +0000 (11:15 -0700)]
Merge tag 'edac_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/ras/ras

Pull EDAC fixes from Borislav Petkov:

 - Fix more issues in the AMD FMPM driver

* tag 'edac_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  RAS: Avoid build errors when CONFIG_DEBUG_FS=n
  RAS/AMD/FMPM: Safely handle saved records of various sizes
  RAS/AMD/FMPM: Avoid NULL ptr deref in get_saved_records()

12 months agoMerge tag 'irq_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 18:04:51 +0000 (11:04 -0700)]
Merge tag 'irq_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Borislav Petkov:

 - Fix an unused function warning on irqchip/irq-armada-370-xp

 - Fix the IRQ sharing with pinctrl-amd and ACPI OSL

* tag 'irq_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/armada-370-xp: Suppress unused-function warning
  genirq: Introduce IRQF_COND_ONESHOT and use it in pinctrl-amd

12 months agoMerge tag 'perf_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 17:43:11 +0000 (10:43 -0700)]
Merge tag 'perf_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/tip/tip

Pull x86 perf fixes from Borislav Petkov:

 - Define the correct set of default hw events on AMD Zen4

 - Use the correct stalled cycles PMCs on AMD Zen2 and newer

 - Fix detection of the LBR freeze feature on AMD

* tag 'perf_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/amd/core: Define a proper ref-cycles event for Zen 4 and later
  perf/x86/amd/core: Update and fix stalled-cycles-* events for Zen 2 and later
  perf/x86/amd/lbr: Use freeze based on availability
  x86/cpufeatures: Add new word for scattered features

12 months agoMerge tag 'timers_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 17:34:49 +0000 (10:34 -0700)]
Merge tag 'timers_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/tip/tip

Pull timers update from Borislav Petkov:

 - Volunteer in Anna-Maria and Frederic as timers co-maintainers so that
   tglx can relax more :-P

* tag 'timers_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  MAINTAINERS: Add co-maintainers for time[rs]

12 months agoMerge tag 'objtool_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 17:30:06 +0000 (10:30 -0700)]
Merge tag 'objtool_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/tip/tip

Pull objtool fix from Borislav Petkov:

 - Fix a format specifier build error in objtool during an x32 build

* tag 'objtool_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  objtool: Fix compile failure when using the x32 compiler

12 months agoMerge tag 'x86_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 31 Mar 2024 17:16:34 +0000 (10:16 -0700)]
Merge tag 'x86_urgent_for_v6.9_rc2' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:

 - Make sure single object builds in arch/x86/virt/ ala
      make ... arch/x86/virt/vmx/tdx/seamcall.o
   work again

 - Do not do ROM range scans and memory validation when the kernel is
   running as a SEV-SNP guest as those can get problematic and, before
   that, are not really needed in such a guest

 - Exclude the build-time generated vdso-image-x32.o object from objtool
   validation and in particular the return sites in there due to a
   warning which fires when an unpatched return thunk is being used

 - Improve the NMI CPUs stall message to show additional information
   about the state of each CPU wrt the NMI handler

 - Enable gcc named address spaces support only on !KCSAN configs due to
   compiler options incompatibility

 - Revert a change which was trying to use GB pages for mapping regions
   only when the regions would be large enough but that change lead to
   kexec failing

 - A documentation fixlet

* tag 'x86_urgent_for_v6.9_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/build: Use obj-y to descend into arch/x86/virt/
  x86/sev: Skip ROM range scans and validation for SEV-SNP guests
  x86/vdso: Fix rethunk patching for vdso-image-x32.o too
  x86/nmi: Upgrade NMI backtrace stall checks & messages
  x86/percpu: Disable named address spaces for KCSAN
  Revert "x86/mm/ident_map: Use gbpages only where full GB page should be mapped."
  Documentation/x86: Fix title underline length

12 months agokconfig: Fix typo HEIGTH to HEIGHT
Isak Ellmer [Sat, 30 Mar 2024 15:19:45 +0000 (16:19 +0100)]
kconfig: Fix typo HEIGTH to HEIGHT

Fixed a typo in some variables where height was misspelled as heigth.

Signed-off-by: Isak Ellmer <isak01@gmail.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
12 months agoDocumentation/llvm: Note s390 LLVM=1 support with LLVM 18.1.0 and newer
Nathan Chancellor [Wed, 27 Mar 2024 17:20:36 +0000 (10:20 -0700)]
Documentation/llvm: Note s390 LLVM=1 support with LLVM 18.1.0 and newer

As of the first s390 pull request during the 6.9 merge window,
commit 691632f0e869 ("Merge tag 's390-6.9-1' of
git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux"), s390 can be
built with LLVM=1 when using LLVM 18.1.0, which is the first version
that has SystemZ support implemented in ld.lld and llvm-objcopy.

Update the supported architectures table in the Kbuild LLVM
documentation to note this explicitly to make it more discoverable by
users and other developers. Additionally, this brings s390 in line with
the rest of the architectures in the table, which all support LLVM=1.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
12 months agokbuild: Disable KCSAN for autogenerated *.mod.c intermediaries
Borislav Petkov (AMD) [Tue, 26 Mar 2024 20:25:48 +0000 (21:25 +0100)]
kbuild: Disable KCSAN for autogenerated *.mod.c intermediaries

When KCSAN and CONSTRUCTORS are enabled, one can trigger the

  "Unpatched return thunk in use. This should not happen!"

catch-all warning.

Usually, when objtool runs on the .o objects, it does generate a section
.return_sites which contains all offsets in the objects to the return
thunks of the functions present there. Those return thunks then get
patched at runtime by the alternatives.

KCSAN and CONSTRUCTORS add this to the object file's .text.startup
section:

  -------------------
  Disassembly of section .text.startup:

  ...

  0000000000000010 <_sub_I_00099_0>:
    10:   f3 0f 1e fa             endbr64
    14:   e8 00 00 00 00          call   19 <_sub_I_00099_0+0x9>
                          15: R_X86_64_PLT32      __tsan_init-0x4
    19:   e9 00 00 00 00          jmp    1e <__UNIQUE_ID___addressable_cryptd_alloc_aead349+0x6>
                          1a: R_X86_64_PLT32      __x86_return_thunk-0x4
  -------------------

which, if it is built as a module goes through the intermediary stage of
creating a <module>.mod.c file which, when translated, receives a second
constructor:

  -------------------
  Disassembly of section .text.startup:

  0000000000000010 <_sub_I_00099_0>:
    10:   f3 0f 1e fa             endbr64
    14:   e8 00 00 00 00          call   19 <_sub_I_00099_0+0x9>
                          15: R_X86_64_PLT32      __tsan_init-0x4
    19:   e9 00 00 00 00          jmp    1e <_sub_I_00099_0+0xe>
                          1a: R_X86_64_PLT32      __x86_return_thunk-0x4

  ...

  0000000000000030 <_sub_I_00099_0>:
    30:   f3 0f 1e fa             endbr64
    34:   e8 00 00 00 00          call   39 <_sub_I_00099_0+0x9>
                          35: R_X86_64_PLT32      __tsan_init-0x4
    39:   e9 00 00 00 00          jmp    3e <__ksymtab_cryptd_alloc_ahash+0x2>
                          3a: R_X86_64_PLT32      __x86_return_thunk-0x4
  -------------------

in the .ko file.

Objtool has run already so that second constructor's return thunk cannot
be added to the .return_sites section and thus the return thunk remains
unpatched and the warning rightfully fires.

Drop KCSAN flags from the mod.c generation stage as those constructors
do not contain data races one would be interested about.

Debugged together with David Kaplan <David.Kaplan@amd.com> and Nikolay
Borisov <nik.borisov@suse.com>.

Reported-by: Paul Menzel <pmenzel@molgen.mpg.de>
Closes: https://lore.kernel.org/r/0851a207-7143-417e-be31-8bf2b3afb57d@molgen.mpg.de
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Tested-by: Paul Menzel <pmenzel@molgen.mpg.de> # Dell XPS 13
Reviewed-by: Nikolay Borisov <nik.borisov@suse.com>
Reviewed-by: Marco Elver <elver@google.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
12 months agokbuild: make -Woverride-init warnings more consistent
Arnd Bergmann [Tue, 26 Mar 2024 14:47:16 +0000 (15:47 +0100)]
kbuild: make -Woverride-init warnings more consistent

The -Woverride-init warn about code that may be intentional or not,
but the inintentional ones tend to be real bugs, so there is a bit of
disagreement on whether this warning option should be enabled by default
and we have multiple settings in scripts/Makefile.extrawarn as well as
individual subsystems.

Older versions of clang only supported -Wno-initializer-overrides with
the same meaning as gcc's -Woverride-init, though all supported versions
now work with both. Because of this difference, an earlier cleanup of
mine accidentally turned the clang warning off for W=1 builds and only
left it on for W=2, while it's still enabled for gcc with W=1.

There is also one driver that only turns the warning off for newer
versions of gcc but not other compilers, and some but not all the
Makefiles still use a cc-disable-warning conditional that is no
longer needed with supported compilers here.

Address all of the above by removing the special cases for clang
and always turning the warning off unconditionally where it got
in the way, using the syntax that is supported by both compilers.

Fixes: 2cd3271b7a31 ("kbuild: avoid duplicate warning options")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Andrew Jeffery <andrew@codeconstruct.com.au>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
12 months agoobjtool: Fix compile failure when using the x32 compiler
Mikulas Patocka [Sat, 30 Mar 2024 19:23:08 +0000 (20:23 +0100)]
objtool: Fix compile failure when using the x32 compiler

When compiling the v6.9-rc1 kernel with the x32 compiler, the following
errors are reported. The reason is that we take an "unsigned long"
variable and print it using "PRIx64" format string.

In file included from check.c:16:
check.c: In function ‘add_dead_ends’:
/usr/src/git/linux-2.6/tools/objtool/include/objtool/warn.h:46:17: error: format ‘%llx’ expects argument of type ‘long long unsigned int’, but argument 5 has type ‘long unsigned int’ [-Werror=format=]
   46 |                 "%s: warning: objtool: " format "\n",   \
      |                 ^~~~~~~~~~~~~~~~~~~~~~~~
check.c:613:33: note: in expansion of macro ‘WARN’
  613 |                                 WARN("can't find unreachable insn at %s+0x%" PRIx64,
      |                                 ^~~~
...

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: linux-kernel@vger.kernel.org
12 months agoMerge tag 'xfs-6.9-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 30 Mar 2024 20:51:58 +0000 (13:51 -0700)]
Merge tag 'xfs-6.9-fixes-1' of git://git./fs/xfs/xfs-linux

Pull xfs fixes from Chandan Babu:

 - Allow stripe unit/width value passed via mount option to be written
   over existing values in the super block

 - Do not set current->journal_info to avoid its value from being miused
   by another filesystem context

* tag 'xfs-6.9-fixes-1' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: don't use current->journal_info
  xfs: allow sunit mount option to repair bad primary sb stripe values