In case of a monitor call program check the CPU stores the monitor code to
lowcore. Let the program check handler copy it to the pt_regs structure so
it can be used by the monitor call exception handler.
Instead of increasing the pt_regs size add a union which contains both
orig_gpr2 and monitor_code, since orig_gpr2 is not used in case of a
program check.
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
The first operand address of the monitor call (mc) instruction is the
monitor code. Currently the monitor code is ignored, but this will
change. Therefore add and use MONCODE_BUG instead of a hardcoded zero.
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
This is just the s390 variant of commit 4f1b701f24 ("x86/bug: Use
BUG_FORMAT for DEBUG_BUGVERBOSE_DETAILED").
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Rewrite the bug inline assembly so it uses input operands again instead of
pure macro replacements. This more or less reverts the conversion done when
'cond_str' support was added [1].
Reason for this is that the upcoming __WARN_printf() implementation
requires an inline assembly with an output operand. At the same time input
strings (format specifier and condition string) may contain the special '%'
character. As soon as an inline assembly is specified to have input/output
operands the '%' has a special meaning: e.g. converting the existing
#define __BUG_FLAGS(cond_str, flags) \
asm_inline volatile(__stringify(ASM_BUG_FLAGS(cond_str, flags)));
to
#define __BUG_FLAGS(cond_str, flags) \
asm_inline volatile(__stringify(ASM_BUG_FLAGS(cond_str, flags))::);
will result in a compile error as soon as 'cond_str' contains a '%'
character:
net/core/neighbour.c: In function ‘neigh_table_init’:
././include/linux/compiler_types.h:546:20: error: invalid 'asm': invalid %-code
...
net/core/neighbour.c:1838:17: note: in expansion of macro ‘WARN_ON’
1838 | WARN_ON(tbl->entry_size % NEIGH_PRIV_ALIGN);
| ^~~~~~~
Convert the code, use immediate operands, and also add comments similar to
x86 which are emitted to the generated assembly file, which makes debugging
much easier.
Note: since gcc-8 does not support strings as immediate input operands,
guard the new implementation with CC_HAS_ASM_IMMEDIATE_STRINGS and fallback
to the generic non-exception based warning implementation for incompatible
compilers.
[1] 6584ff203a ("bugs/s390: Use 'cond_str' in __EMIT_BUG()")
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Upcoming changes to s390 specific inline assemblies require the usage of
strings as immediate input operands. This works only with gcc-9 and newer
compilers. With gcc-8 this leads to a compile error:
void bar(void) { asm volatile("" :: "i" ("foo")); }
Results in:
In function 'bar':
warning: asm operand 0 probably doesn't match constraints
asm volatile("" :: "i" ("foo"));
^~~
error: impossible constraint in 'asm'
Provide a CC_HAS_ASM_IMMEDIATE_STRINGS config option which allows to tell
if the compiler supports strings as immediate input operands.
Based on that conditional code can be provided.
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
While pci_read() and pci_write() have returned errno values since their
inception with commit cd24834130 ("s390/pci: base support") other
config accessors in particular pci_generic_config_read() as well as
pci_generic_config_write() return specific error values which are then
converted to errno by pcibios_err_to_errno(). Since latter does handle
the case where the error value already looks like an errno the previous
behavior is unlikely to cause actual issues.
Still, for consistency and in case any caller explicitly checks error
values align pci_read() and pci_write() with the generic accessors.
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Reviewed-by: Farhan Ali <alifm@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
On s390 PCI busses and slots with multiple functions may have holes
because PCI functions are passed-through by the hypervisor on a per
function basis and some functions may be in standby or reserved. This
fact is indicated by returning true from the
hypervisor_isolated_pci_functions() helper and triggers common code to
scan all possible devfn values. Via pci_scan_single_device() this in
turn causes config reads for the device and vendor IDs, even for PCI
functions which are in standby and thereofore disabled.
So far these futile config reads, as well as potentially writes, which
can never succeed were handled by the PCI load/store instructions
themselves. This works as the platform just returns an error for
a disabled and thus not usable function handle. It does cause spamming
of error logs and additional overhead though.
Instead check if the used function handle is enabled in zpci_cfg_load()
and zpci_cfg_write() and if not enable directly return -ENODEV. Also
refactor zpci_cfg_load() and zpci_cfg_store() slightly to accommodate
the new logic while meeting modern kernel style guidelines.
Cc: stable@vger.kernel.org
Fixes: a50297cf82 ("s390/pci: separate zbus creation from scanning")
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Reviewed-by: Farhan Ali <alifm@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Heiko Carstens says:
====================
The option to select PREEMPT_NONE will go away for all architectures which
support PREEMPT_LAZY [1]. Until now all distributions provide kernels built
with PREEMPT_NONE enabled for s390. In particular this means that all
preempt_disable() / preempt_enable() pairs are optimized away during
compile time.
With PREEMPT_LAZY this is not the case. Switching to PREEMPT_LAZY leads
to a kernel image size increase of ~218kb (defconfig, gcc15).
s390 provides optimized preempt primitives, however there is still room for
improvement. Since support for relocatable lowcore was added access to
preempt_count in lowcore requires an extra call of get_lowcore(), which
generates an extra instruction. Also all instructions have to use a base
register which is not zero to access preempt_count.
Address this by adding a couple of inline assemblies with alternatives.
This generates better code and reduces the size of a PREEMPT_LAZY built
kernel image by ~58kb.
[1] https://lore.kernel.org/all/20251219101502.GB1132199@noisy.programming.kicks-ass.net/
====================
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Provide an inline assembly using alternatives to avoid the need of a
base register due to relocatable lowcore when adding or subtracting
small constants from preempt_count.
Main user is preempt_enable(), which subtracts one from preempt_count
and tests if the result is zero.
With this the generated code changes from
1000b8: a7 19 00 00 lghi %r1,0
1000bc: eb ff 13 a8 00 6e alsi 936(%r1),-1
1000c2: a7 54 00 05 jnhe 1000cc <__rcu_read_unlock+0x14>
to something like this:
1000b8: eb ff 03 a8 00 6e alsi 936,-1
1000be: a7 54 00 05 jnhe 1000c8 <__rcu_read_unlock+0x10>
Kernel image size is reduced by 45kb (bloat-o-meter -t, defconfig, gcc15).
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
With the empty define __is_enabled(__HAVE_ASM_FLAG_OUTPUTS__)
evaluates to false. Therefore let __HAVE_ASM_FLAG_OUTPUTS__ define 1
if it is defined.
This allows to make use of __is_defined(__HAVE_ASM_FLAG_OUTPUTS__)
like expected.
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Provide an inline assembly using alternatives to avoid the need of a
base register due to relocatable lowcore when adding or subtracting
small constants from preempt_count.
Main user is preempt_disable(), which subtracts one from
preempt_count. With this the generated code changes from
10012c: a7 b9 00 00 lghi %r11,0
100130: eb 01 b3 a8 00 6a asi 936(%r11),1
to something like this:
10012c: eb 01 03 a8 00 6a asi 936,1
Kernel image size is reduced by 13kb (bloat-o-meter -t, defconfig, gcc15).
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Provide an inline assembly using alternatives to avoid the need of
a base register when reading preempt_count() from lowcore. Use the
LLGT instruction, which reads only the least significant 31 bits of
preempt_count. This masks out the encoded PREEMPT_NEED_RESCHED bit.
Generated code is changed from
000000000046e5d0 <vfree>:
46e5d0: c0 04 00 00 00 00 jgnop 46e5d0 <vfree>
46e5d6: a7 39 00 00 lghi %r3,0
46e5da: 58 10 33 a8 l %r1,936(%r3)
46e5de: c0 1b 00 ff ff 00 nilf %r1,16776960
46e5e4: a7 74 00 11 jne 46e606 <vfree+0x36>
to something like this:
000000000046e5d0 <vfree>:
46e5d0: c0 04 00 00 00 00 jgnop 46e5d0 <vfree>
46e5d6: e3 10 03 a8 00 17 llgt %r1,936
46e5dc: ec 41 28 b7 00 55 risbgz %r4,%r1,40,55
46e5e2: a7 74 00 0f jne 46e600 <vfree+0x30>
Overall savings are only 82 bytes according to bloat-o-meter. This
is because of different inlining decisions, and there aren't many
preempt_count() users in the kernel.
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
The generic CC_CAN_LINK detection relies on -m32/-m64 compiler flags.
Some s390 toolchains use -m31 instead but that is not supported in the
kernel.
Make the logic easier to understand and allow the simplification of the
generic CC_CAN_LINK by using a tailored implementation.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Revisit and rework module parameter permissions for AP bus and zcrypt
device drivers.
In general all sysfs permissions for AP bus and zcrypt parameters should be
0444 so that user space tools like lszcrypt can read the current value of
module parameters.
Some exceptions are only for some internal tweak parameters like
ap_msg_pool_min_items and zcrypt_mempool_threshold which should only be
readable by an administrator.
Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
Reviewed-by: Holger Dengler <dengler@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Convert the function macros user_mode(), instruction_pointer(), and
user_stack_pointer() to inline functions, to align their definition
with x86 and arm64.
Use const qualifier on struct pt_regs pointer parameters to prevent
compiler warnings:
arch/s390/kernel/stacktrace.c: In function ‘arch_stack_walk_user_common’:
arch/s390/kernel/stacktrace.c:114:34: warning: passing argument 1 of
‘instruction_pointer’ discards ‘const’ qualifier from pointer target
type [-Wdiscarded-qualifiers]
...
arch/s390/kernel/stacktrace.c:117:48: warning: passing argument 1 of
‘user_stack_pointer’ discards ‘const’ qualifier from pointer target
type [-Wdiscarded-qualifiers]
...
While at it add const qualifier to all struct pt_regs pointer parameters
that are accessed read-only and use __always_inline instead of inline to
harmonize the helper functions.
[hca@linux.ibm.com: Use psw_bits() helper]
Reviewed-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Add -Wno-default-const-init-unsafe to purgatory KBUILD_CFLAGS, similar
to scripts/Makefile.extrawarn, since clang generates warnings for the
dummy variable in typecheck():
CC arch/s390/purgatory/purgatory.o
arch/s390/include/asm/ptrace.h:221:9: warning: default initialization of an object of type 'typeof (regs->psw)' (aka 'const psw_t') leaves the object uninitialized [-Wdefault-const-init-var-unsafe]
221 | return psw_bits(regs->psw).pstate;
| ^
arch/s390/include/asm/ptrace.h:98:2: note: expanded from macro 'psw_bits'
98 | typecheck(psw_t, __psw); \
| ^
include/linux/typecheck.h:11:12: note: expanded from macro 'typecheck'
11 | typeof(x) __dummy2; \
| ^
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Add -Wno-default-const-init-unsafe to boot KBUILD_CFLAGS, similar to
scripts/Makefile.extrawarn, since clang generates warnings for the dummy
variable in typecheck():
CC arch/s390/boot/version.o
arch/s390/include/asm/ptrace.h:221:9: warning: default initialization of an object of type 'typeof (regs->psw)' (aka 'const psw_t') leaves the object uninitialized [-Wdefault-const-init-var-unsafe]
221 | return psw_bits(regs->psw).pstate;
| ^
arch/s390/include/asm/ptrace.h:98:2: note: expanded from macro 'psw_bits'
98 | typecheck(psw_t, __psw); \
| ^
include/linux/typecheck.h:11:12: note: expanded from macro 'typecheck'
11 | typeof(x) __dummy2; \
| ^
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Pull Coccinelle fixes from Julia Lawall:
"These fix a typo and make the coccicheck script more robust by
ensuring that only compatible semantic patches are executed for the
chosen mode"
* tag 'coccinelle-6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/jlawall/linux:
Coccinelle: pm_runtime: Fix typo in report message
scripts: coccicheck: filter *.cocci files by MODE
Pull input fixes from Dmitry Torokhov:
- a quirk for i8042 to better handle another TUXEDO model
- a quirk to atkbd to handle incorcet behavior of HONOR FMB-P internal
keyboard
- a definition for a new ABS_SND_PROFILE event
- fixes to alps and lkkbd drivers to reliably shut down pending work on
removal
- a fix to apple_z2 driver tightening input report parsing
- a fix for "off-by-one" error when validating config in ti_am335x_tsc
driver
- addition of CRKD Guitars device IDs to xpad driver.
* tag 'input-for-v6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: ti_am335x_tsc - fix off-by-one error in wire_order validation
Input: xpad - add support for CRKD Guitars
Input: add ABS_SND_PROFILE
Input: apple_z2 - fix reading incorrect reports after exiting sleep
Input: alps - fix use-after-free bugs caused by dev3_register_work
Input: i8042 - add TUXEDO InfinityBook Max Gen10 AMD to i8042 quirk table
Input: atkbd - skip deactivate for HONOR FMB-P's internal keyboard
Input: lkkbd - disable pending work before freeing device
Pull x86 fixes from Ingo Molnar:
- Fix FPU core dumps on certain CPU models
- Fix htmldocs build warning
- Export TLB tracing event name via header
- Remove unused constant from <linux/mm_types.h>
- Fix comments
- Fix whitespace noise in documentation
- Fix variadic structure's definition to un-confuse UBSAN
- Fix posted MSI interrupts irq_retrigger() bug
- Fix asm build failure with older GCC builds
* tag 'x86-urgent-2025-12-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/bug: Fix old GCC compile fails
x86/msi: Make irq_retrigger() functional for posted MSI
x86/platform/uv: Fix UBSAN array-index-out-of-bounds
mm: Remove tlb_flush_reason::NR_TLB_FLUSH_REASONS from <linux/mm_types.h>
x86/mm/tlb/trace: Export the TLB_REMOTE_WRONG_CPU enum in <trace/events/tlb.h>
x86/sgx: Remove unmatched quote in __sgx_encl_extend function comment
x86/boot/Documentation: Fix whitespace noise in boot.rst
x86/fpu: Fix FPU state core dump truncation on CPUs with no extended xfeatures
x86/boot/Documentation: Fix htmldocs build warning due to malformed table in boot.rst
Enhance the coccicheck script to filter *.cocci files based on the
specified MODE (e.g., report, patch). This ensures that only compatible
semantic patch files are executed, preventing errors such as:
"virtual rule report not supported"
This error occurs when a .cocci file does not define a 'virtual <MODE>'
rule, yet is executed in that mode.
For example:
make coccicheck M=drivers/hwtracing/coresight/ MODE=report
In this case, running "secs_to_jiffies.cocci" would trigger the error
because it lacks support for 'report' mode. With this change, such files
are skipped automatically, improving robustness and developer
experience.
Signed-off-by: Songwei Chai <quic_songchai@quicinc.com>
Reviewed-by: Julia Lawall <Julia.Lawall@inria.fr>
Pull ata fix from Damien Le Moal:
- Disable link power management (LPM) for a Seagate drive that is
misbehaving when LPM is enabled
* tag 'ata-6.19-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux:
ata: libata-core: Disable LPM on ST2000DM008-2FR102
Pull spi fixes from Mark Brown:
"A small collection of fixes for various SPI drivers, plus a relaxation
of constraints in the DT for the DesignWare controller to reflect
hardware that's been seen.
There's several fixes for the Cadence QuadSPI driver since a fix
during the last release made some existing issues with error handling
during probe more readily visible"
* tag 'spi-fix-v6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
spi: mt65xx: Use IRQF_ONESHOT with threaded IRQ
spi: dt-bindings: snps,dw-abp-ssi: Allow up to 16 chip-selects
spi: cadence-quadspi: Fix clock disable on probe failure path
spi: cadence-quadspi: Add error logging for DMA request failure
spi: fsl-cpm: Check length parity before switching to 16 bit mode
spi: mpfs: Fix an error handling path in mpfs_spi_probe()
clang is generating very inefficient code for native_save_fl() which is
used for local_irq_save() in critical spots.
Allowing the "pop %0" to use memory:
1) forces the compiler to add annoying stack canaries when
CONFIG_STACKPROTECTOR_STRONG=y in many places.
2) Almost always is followed by an immediate "move memory,register"
One good example is _raw_spin_lock_irqsave, with 8 extra instructions
ffffffff82067a30 <_raw_spin_lock_irqsave>:
ffffffff82067a30: ...
ffffffff82067a39: 53 push %rbx
// Three instructions to ajust the stack, read the per-cpu canary
// and copy it to 8(%rsp)
ffffffff82067a3a: 48 83 ec 10 sub $0x10,%rsp
ffffffff82067a3e: 65 48 8b 05 da 15 45 02 mov %gs:0x24515da(%rip),%rax # <__stack_chk_guard>
ffffffff82067a46: 48 89 44 24 08 mov %rax,0x8(%rsp)
ffffffff82067a4b: 9c pushf
// instead of pop %rbx, compiler uses 2 instructions.
ffffffff82067a4c: 8f 04 24 pop (%rsp)
ffffffff82067a4f: 48 8b 1c 24 mov (%rsp),%rbx
ffffffff82067a53: fa cli
ffffffff82067a54: b9 01 00 00 00 mov $0x1,%ecx
ffffffff82067a59: 31 c0 xor %eax,%eax
ffffffff82067a5b: f0 0f b1 0f lock cmpxchg %ecx,(%rdi)
ffffffff82067a5f: 75 1d jne ffffffff82067a7e <_raw_spin_lock_irqsave+0x4e>
// three instructions to check the stack canary
ffffffff82067a61: 65 48 8b 05 b7 15 45 02 mov %gs:0x24515b7(%rip),%rax # <__stack_chk_guard>
ffffffff82067a69: 48 3b 44 24 08 cmp 0x8(%rsp),%rax
ffffffff82067a6e: 75 17 jne ffffffff82067a87
...
// One extra instruction to adjust the stack.
ffffffff82067a73: 48 83 c4 10 add $0x10,%rsp
...
// One more instruction in case the stack was mangled.
ffffffff82067a87: e8 a4 35 ff ff call ffffffff8205b030 <__stack_chk_fail>
This patch changes nothing for gcc, but for clang saves ~20000 bytes of text
even though more functions are inlined.
$ size vmlinux.gcc.before vmlinux.gcc.after vmlinux.clang.before vmlinux.clang.after
text data bss dec hex filename
45565821 25005462 4704800 75276083 47c9f33 vmlinux.gcc.before
45565821 25005462 4704800 75276083 47c9f33 vmlinux.gcc.after
45121072 24638617 5533040 75292729 47ce039 vmlinux.clang.before
45093887 24638633 5536808 75269328 47c84d0 vmlinux.clang.after
$ scripts/bloat-o-meter -t vmlinux.clang.before vmlinux.clang.after
add/remove: 1/2 grow/shrink: 21/533 up/down: 2250/-22112 (-19862)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Work around clang problems with "=rm" asm constraint.
clang seems to always chose the memory output, while it is almost
always the worst choice.
Add ASM_OUTPUT_RM so that we can replace "=rm" constraint
where it matters for clang, while not penalizing gcc.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Suggested-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull xfs fixes from Carlos Maiolino:
"This contains a few fixes for zoned devices support, an UAF and a
compiler warning, and some cleaning up"
* tag 'xfs-fixes-6.19-rc2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: fix the zoned RT growfs check for zone alignment
xfs: validate that zoned RT devices are zone aligned
xfs: fix XFS_ERRTAG_FORCE_ZERO_RANGE for zoned file system
xfs: fix a memory leak in xfs_buf_item_init()
xfs: fix stupid compiler warning
xfs: fix a UAF problem in xattr repair
xfs: ignore discard return value
Pull drm fixes from Dave Airlie:
"rc2 fixes for the week, mostly xe, with amdgpu as usual. Then a
smattering of small fixes across the core/tests/panel and amdxdna.
I expect things will be quiet for rc3/4 as teams take a break, and I'm
travelling but will keep an eye on things.
core:
- fix gem handle leak on DRM_IOCTL_GEM_CHANGE_HANDLE
tests:
- add EDEADLK handling
amdgpu:
- Fix no_console_suspend handling
- DCN 3.5.x seamless boot fixes
- DP audio fix
- Fix race in GPU recovery
- SMU 14 OD fix
amdkfd:
- Event fix
xe:
- Limit num_syncs to prevent oversized kernel allocations
- Disallow 0 OA property values
- Disallow 0 EU stall property values
- Fix kobject leak
- Workaround
- Loop variable reference fix
- Fix a CONFIG corner-case incorrect number of argument
- Skip reason prefix while emitting array
- VF migration fix
- Fix context in mei interrupt top half
- Don't include the CCS metadata in the dma-buf sg-table
- VF queueing recovery work fix
- Increase TDF timeout
- GT reset registers vs scheduler ordering fix
- Adjust long-running workload timeslices
- Always set OA_OAGLBCTXCTRL_COUNTER_RESUME
- Fix a return value
- Drop preempt-fences when destroying imported dma-bufs
- Use usleep_range for accurate long-running workload timeslicing
amdxdna:
- don't load virtualized
panel:
- fix visionox-rm69299 Kconfig dependency
- sony-td4353-jdi probing fix"
* tag 'drm-fixes-2025-12-20' of https://gitlab.freedesktop.org/drm/kernel: (34 commits)
drm/xe: Use usleep_range for accurate long-running workload timeslicing
drm/xe: Drop preempt-fences when destroying imported dma-bufs.
drm/xe/eustall: Disallow 0 EU stall property values
drm/xe/oa: Disallow 0 OA property values
drm/xe/xe_sriov_vfio: Fix return value in xe_sriov_vfio_migration_supported()
drm/xe/oa: Always set OAG_OAGLBCTXCTRL_COUNTER_RESUME
drm/xe: Adjust long-running workload timeslices to reasonable values
drm/xe/oa: Limit num_syncs to prevent oversized allocations
drm/xe: Limit num_syncs to prevent oversized allocations
drm/amdkfd: Fix improper NULL termination of queue restore SMI event string
drm/amd/pm: restore SCLK settings after S0ix resume
drm/amdgpu: fix a job->pasid access race in gpu recovery
drm/amd/display: Fix DP no audio issue
drm/amd/display: Fix scratch registers offsets for DCN351
drm/amd/display: Fix scratch registers offsets for DCN35
drm/amd: Resume the device in thaw() callback when console suspend is disabled
drm/panel: visionox-rm69299: Depend on BACKLIGHT_CLASS_DEVICE
accel/amdxdna: Block running under a hypervisor
drm/panel: sony-td4353-jdi: Enable prepare_prev_first
drm/xe: Restore engine registers before restarting schedulers after GT reset
...
Pull kunit fixes from Shuah Khan:
"Drop unused parameter from kunit_device_register_internal and make
FAULT_TEST default to n when PANIC_ON_OOPS"
* tag 'linux_kselftest-kunit-fixes-6.19-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
kunit: make FAULT_TEST default to n when PANIC_ON_OOPS
kunit: Drop unused parameter from kunit_device_register_internal
Pull MIPS fixes from Thomas Bogendoerfer:
- Fix build error for Alchemy
- Fix reference leak
* tag 'mips-fixes_6.19_1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
MIPS: Fix a reference leak bug in ip22_check_gio()
MIPS: Alchemy: Remove bogus static/inline specifiers
Pull arm64 fixes from Catalin Marinas:
"Two left-over updates that could not go into -rc1 due to conflicts
with other series:
- Simplify checks in arch_kfence_init_pool() since
force_pte_mapping() already takes BBML2-noabort (break-before-make
Level 2 with no aborts generated) into account
- Remove unneeded SVE/SME fallback preserve/store handling in the
arm64 EFI. With the recent updates, the fallback path is only taken
for EFI runtime calls from hardirq or NMI contexts. In practice,
this only happens under panic/oops/emergency_restart() and no
restoring of the user state expected.
There's a corresponding lkdtm update to trigger a BUG() or panic()
from hardirq context together with a fixup not to confuse
clang/objtool about the control flow
GCS (guarded control stacks) fix: flush the GCS locking state on exec,
otherwise the new task will not be able to enable GCS (locked as
disabled)"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
lkdtm/bugs: Do not confuse the clang/objtool with busy wait loop
arm64/gcs: Flush the GCS locking state on exec
arm64/efi: Remove unneeded SVE/SME fallback preserve/store handling
lkdtm/bugs: Add cases for BUG and PANIC occurring in hardirq context
arm64: mm: Simplify check in arch_kfence_init_pool()
Pull x86 kvm fixes from Paolo Bonzini:
"x86 fixes. Everyone else is already in holiday mood apparently.
- Add a missing 'break' to fix param parsing in the rseq selftest
- Apply runtime updates to the _current_ CPUID when userspace is
setting CPUID, e.g. as part of vCPU hotplug, to fix a false
positive and to avoid dropping the pending update
- Disallow toggling KVM_MEM_GUEST_MEMFD on an existing memslot, as
it's not supported by KVM and leads to a use-after-free due to KVM
failing to unbind the memslot from the previously-associated
guest_memfd instance
- Harden against similar KVM_MEM_GUEST_MEMFD goofs, and prepare for
supporting flags-only changes on KVM_MEM_GUEST_MEMFD memlslots,
e.g. for dirty logging
- Set exit_code[63:32] to -1 (all 0xffs) when synthesizing a nested
SVM_EXIT_ERR (a.k.a. VMEXIT_INVALID) #VMEXIT, as VMEXIT_INVALID is
defined as -1ull (a 64-bit value)
- Update SVI when activating APICv to fix a bug where a
post-activation EOI for an in-service IRQ would effective be lost
due to SVI being stale
- Immediately refresh APICv controls (if necessary) on a nested
VM-Exit instead of deferring the update via KVM_REQ_APICV_UPDATE,
as the request is effectively ignored because KVM thinks the vCPU
already has the correct APICv settings"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: nVMX: Immediately refresh APICv controls as needed on nested VM-Exit
KVM: VMX: Update SVI during runtime APICv activation
KVM: nSVM: Set exit_code_hi to -1 when synthesizing SVM_EXIT_ERR (failed VMRUN)
KVM: nSVM: Clear exit_code_hi in VMCB when synthesizing nested VM-Exits
KVM: Harden and prepare for modifying existing guest_memfd memslots
KVM: Disallow toggling KVM_MEM_GUEST_MEMFD on an existing memslot
KVM: selftests: Add a CPUID testcase for KVM_SET_CPUID2 with runtime updates
KVM: x86: Apply runtime updates to current CPUID during KVM_SET_CPUID{,2}
KVM: selftests: Add missing "break" in rseq_test's param parsing
Pull xen fix from Juergen Gross:
"Just a single patch fixing a sparse warning"
* tag 'for-linus-6.19-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
x86/xen: Fix sparse warning in enlighten_pv.c
Pull slab fix from Vlastimil Babka:
- A stable fix for a missing tag reset that can happen in
kfree_nolock() with KASAN+SLUB_TINY configs (Deepanshu Kartikey)
* tag 'slab-for-6.19-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/vbabka/slab:
mm/slub: reset KASAN tag in defer_free() before accessing freed memory
Pull iommu fixes from Joerg Roedel:
- iommupt: Fix an oops found by syzcaller in the new generic
IO-page-table code.
- AMD-Vi: Fix IO_PAGE_FAULTs in kdump kernels triggered by re-using
domain-ids from previous kernel.
* tag 'iommu-fixes-v6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux:
amd/iommu: Make protection domain ID functions non-static
amd/iommu: Preserve domain ids inside the kdump kernel
iommupt: Return ERR_PTR from _table_alloc()
Pull block fixes from Jens Axboe:
- ublk selftests for missing coverage
- two fixes for the block integrity code
- fix for the newly added newly added PR read keys ioctl, limiting the
memory that can be allocated
- work around for a deadlock that can occur with ublk, where partition
scanning ends up recursing back into file closure, which needs the
same mutex grabbed. Not the prettiest thing in the world, but an
acceptable work-around until we can eliminate the reliance on
disk->open_mutex for this
- fix for a race between enabling writeback throttling and new IO
submissions
- move a bit of bio flag handling code. No changes, but needed for a
patchset for a future kernel
- fix for an init time id leak failure in rnbd
- loop/zloop state check fix
* tag 'block-6.19-20251218' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux:
block: validate interval_exp integrity limit
block: validate pi_offset integrity limit
block: rnbd-clt: Fix leaked ID in init_dev()
ublk: fix deadlock when reading partition table
block: add allocation size check in blkdev_pr_read_keys()
Documentation: admin-guide: blockdev: replace zone_capacity with zone_capacity_mb when creating devices
zloop: use READ_ONCE() to read lo->lo_state in queue_rq path
loop: use READ_ONCE() to read lo->lo_state without locking
block: fix race between wbt_enable_default and IO submission
selftests: ublk: add user copy test cases
selftests: ublk: add support for user copy to kublk
selftests: ublk: forbid multiple data copy modes
selftests: ublk: don't share backing files between ublk servers
selftests: ublk: use auto_zc for PER_IO_DAEMON tests in stress_04
selftests: ublk: fix fio arguments in run_io_and_recover()
selftests: ublk: remove unused ios map in seq_io.bt
selftests: ublk: correct last_rw map type in seq_io.bt
selftests: ublk: fix overflow in ublk_queue_auto_zc_fallback()
block: move around bio flagging helpers
Pull io_uring fix from Jens Axboe:
"Just a single fix this week, for an issue with the calculation of the
number of segments in the ublk kbuf import path"
* tag 'io_uring-6.19-20251218' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux:
io_uring: fix nr_segs calculation in io_import_kbuf
Since commit eb972eab07 ("lkdtm/bugs: Add cases for BUG and PANIC
occurring in hardirq context"), building with clang for x86_64 results
in the following warnings:
vmlinux.o: warning: objtool: lkdtm_PANIC_IN_HARDIRQ(): unexpected end of section .text.lkdtm_PANIC_IN_HARDIRQ
vmlinux.o: warning: objtool: lkdtm_BUG_IN_HARDIRQ(): unexpected end of section .text.lkdtm_BUG_IN_HARDIRQ
caused by busy "while (wait_for_...);" loops. Add READ_ONCE() and
cpu_relax() to better indicate the intention and avoid any unwanted
compiler optimisations.
Fixes: eb972eab07 ("lkdtm/bugs: Add cases for BUG and PANIC occurring in hardirq context")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202512190111.jxFSqxUH-lkp@intel.com/
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
So that both iommu.c and init.c can utilize them. Also define a new
function 'pdom_id_destroy()' to destroy 'pdom_ids' instead of directly
calling ida functions.
Signed-off-by: Sairaj Kodilkar <sarunkod@amd.com>
Reviewed-by: Vasant Hegde <vasant.hegde@amd.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Heiko Carstens
Niklas Schnelle
Julia Lawall
Thomas Weißschuh
Harald Freudenberger
Jens Remus
Linus Torvalds
Thorsten Blum
Songwei Chai
Eric Dumazet
Nuno Sá
Rob Herring (Arm)
Frank Wunderlich
Catalin Marinas
Sairaj Kodilkar