This reverts commit a345ee7b53.
This causes merge issues with upstream patches and can be reapplied
later if really needed.
Bug: 276339429
Cc: Yifan Hong <elsk@google.com>
Change-Id: I7222970b51dfab52ab78ea4a2203b4d74003dcb8
Signed-off-by: Carlos Llamas <cmllamas@google.com>
This reverts commit 2f0acef0f9.
The headers are no longer generated since commit bf7b376659 ("Revert
"ANDROID: GKI: Add script to generate symbol protection headers""), so
these are no longer needed in gitignore. Plus they cause merge conflicts
with upstream changes.
Change-Id: Iecae47593f619c9c99b65f0837828a023c358d8a
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Upstream renamed this test in commit 0c6f0d77ab ("rtc: Rename lib_test
to test_rtc_lib"), so update our modules.bzl and kunit test configs
accordingly.
Fixes: 0c6f0d77ab ("rtc: Rename lib_test to test_rtc_lib")
Change-Id: I7dbacb6f0a25338782ea5f95fab0655ee998edb1
Signed-off-by: Carlos Llamas <cmllamas@google.com>
This reverts commit 8e717112ca.
This breaks Pixel's drivers that are still using pp->msi_domain, so go
back to the legacy implementation while these drivers are updated to use
the new MSI setup. This patch can be reapplied then.
Bug: 442591113
Change-Id: I7d1075323f64d1a2abeafcecf01335969ad37fab
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Steps on the way to v6.17-rc1
Resolves merge conflicts in:
include/linux/sched.h
Cc: John Stultz <jstultz@google.com>
Change-Id: Idbe526cc9916b717409d5b85f84d15fec3f8f5f3
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Steps on the way to v6.17-rc1
Resolves merge conflicts in:
include/linux/power_supply.h
Change-Id: Ia5db25b3e13965601d52b5ebf1488aefbe5deb85
Signed-off-by: Carlos Llamas <cmllamas@google.com>
memzero_page() is the new name for zero_user().
Cc: Paul Lawrence <paullawrence@google.com>
Fixes: 234dda7a49 ("mm: remove zero_user()")
Change-Id: I21d55988227ef7a18ae643670e455dfed4281fc9
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Convert call to filemap_grab_folio() as grab_cache_page() was removed in
commit 45cd52c44e ("mm: remove grab_cache_page()"). Also convert other
subsequent calls manipulating the page to folio equivalents.
Fixes: 45cd52c44e ("mm: remove grab_cache_page()")
Cc: Paul Lawrence <paullawrence@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Change-Id: I050558b4d14142fe8072284993a1f9b498212512
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Commit 3215eaceca ("mm/mremap: refactor initial parameter sanity
checks") moved the sanity check for vrm->new_addr from mremap_to() to
check_mremap_params().
However, this caused a regression as vrm->new_addr is now checked even
when MREMAP_FIXED and MREMAP_DONTUNMAP flags are not specified. In this
case, vrm->new_addr can be garbage and create unexpected failures.
Fix this by moving the new_addr check after the vrm_implies_new_addr()
guard. This ensures that the new_addr is only checked when the user has
specified one explicitly.
Cc: stable@vger.kernel.org
Fixes: 3215eaceca ("mm/mremap: refactor initial parameter sanity checks")
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Bug: 441615946
Link: https://lore.kernel.org/all/20250828142657.770502-1-cmllamas@google.com/
Change-Id: I29e4326094e463a0acd1d2a206682a0f1f787299
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Steps on the way to v6.17-rc1
Resolves merge conflicts in:
mm/cma.c
mm/page_alloc.c
Cc: Suren Baghdasaryan <surenb@google.com>
Change-Id: Id6ae7e1446bea2a386152f164a4aedf75d335d43
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Upstream commit bde7170a04 ("netfilter: xtables: disable 32bit compat
interface by default") silently disabled
CONFIG_NETFILTER_XTABLES_COMPAT, dropping support for x_tables from
32-bit operating systems.
The GKI kernel is used by devices that still run a 32-bit OS, so enable
that config again.
Bug: 437140699
Change-Id: Ifa4db2af156284c87e7b38106cfad6c898285145
[isaacmanjarres: resolved merge conflicts when porting forward
from android15-6.6.]
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
Required for use of PCI devices in a protected VM.
Bug: 436349788
Test: build (only carefully tested android15-6.6 branch)
Signed-off-by: Frederick Mayle <fmayle@google.com>
Change-Id: Ica333eed62c4ce7ce7e7237300effd219a0071e4
aarch64 was hitting the following error, CONFIG_VMAP_STACK appears to be
forced on for arm64 now:
ERROR: CONFIG_VMAP_STACK: actual 'y', expected '' from common/arch/arm64/configs/crashdump_defconfig
Similarly, x86_64 seems to require CONFIG_SPARSEMEM_VMEMMAP, probably
because of `ARCH_SELECT_MEMORY_MODEL=y` (on by default):
ERROR: CONFIG_SPARSEMEM_VMEMMAP: actual 'y', expected '' from common/arch/x86/configs/crashdump_defconfig
Bug: 436349788
Test: build
Change-Id: I7c0617990c5e398d4568b724fdee5b460ca92228
Signed-off-by: Frederick Mayle <fmayle@google.com>
Add the kcmp kselftest to all architectures and to the VTS
test configuration
Bug: 416599813
Signed-off-by: Wake Liu <wakel@google.com>
Change-Id: I52498b66aee85527209366381b0783d9dad6f801
Add the futex kselftest to all architectures and to the VTS
test configuration.
Bug: 416599813
Signed-off-by: Wake Liu <wakel@google.com>
Change-Id: I1e21ff962484d0c0920a6877a122bf3b9dcd5a51
Add the capabilities kselftest to all architectures and to the VTS
test configuration.
Bug: 416599813
Signed-off-by: Wake Liu <wakel@google.com>
Change-Id: I0b307774da7aac479eda909f8f8cd626ea8c6a3b
Unlike
commit 7a2f90ea8a ("ANDROID: Create list of protected module names"),
this change processes the module name list in the analysis phase so that
we don't have to process it at the macro expansion phase. This allows
future changes to make get_gki_modules_list and get_kunit_modules_list
return a select() expression.
Bug: 393366754
Bug: 419603781
Test: manually build these two targets
Change-Id: Ia0b2fd4dc92c75061fff2f8fba68bbb37bfa7b49
Signed-off-by: HONG Yifan <elsk@google.com>
Unlike the get_gki_modules_list and get_kunit_modules_list,
these new functions get_gki_modules_superset
and get_kunit_modules_superset returns a plain
list with all items in each conditional branch of the
conditional.
Bug: 419603781
Change-Id: I07c3cb5ca876833ebc61010dbf19ba50b5c491b4
Signed-off-by: HONG Yifan <elsk@google.com>
The important thing for cherry-picks FROMGIT are that the content of the
change doesn't change (by the maintainer just before a merge request, or
due to rebasing a tree), not so much that it is a "stable maintainer"
tree, so update the language a little.
Also, explicitly call out 'net' and 'net-next' as these trees are
"special", and in the opposite way to 'linux-next'.
Change-Id: I7b2c009d237fbc44bb515fce497cdd92e2138420
Signed-off-by: Alistair Delva <adelva@google.com>
Enable CONFIG_ZRAM_WRITEBACK, CONFIG_ZRAM_MULTI_COMP (i.e.
recompression), and CONFIG_ZRAM_BACKEND_ZSTD. Enabling these allows the
new mmd system component to manage writeback and recompression when
configured to do so, without the need of providing zram as a vendor
module.
Bug: 404275001
Change-Id: I6ae1fa9768408b1b9782747c98ba02b9c0f7db3e
Signed-off-by: David Stevens <stevensd@google.com>
The CONFIG_ARM64_PSEUDO_NMI configuration macro needs to be enabled. Pseudo NMI has many benefits:
(1) It solves the problem of sensitive interrupts being delayed due to lock interrupts, significantly reducing the response delay of sensitive interrupts
(2) It can help locate hard lockup problems caused by interrupt masking
(3) Through IPI_CPU_STOP_NMI, it can also detect CPU stop failures caused by interrupt masking
There is no disadvantage, because if the user does not want to use the pseudo NMI function, the default setting of irqchip.gicv3_pseudo_nmi is disabled.
Bug: 439502098
Change-Id: Ib19037853a844e62cb13487656720334ed875de4
Signed-off-by: linjunjie1 <linjunjie1@xiaomi.corp-partner.google.com>
In commit 45a2974157 ("drm/msm: Use the central UBWC config database")
the QCOM_UBWC_CONFIG gets selected for config DRM_MSM, so add it to the
db845c modules list.
Fixes: 45a2974157 ("drm/msm: Use the central UBWC config database")
Change-Id: Ie3c3158138747ecc2f7262fb1e24c18b2d54c6fa
Signed-off-by: Carlos Llamas <cmllamas@google.com>
In commit 111fdd2198 ("drm/msm: drm_gpuvm conversion"), the DRM_GPUVM
module gets selected for config DRM_MSM, so add it to the db845c modules
list.
Fixes: 111fdd2198 ("drm/msm: drm_gpuvm conversion")
Change-Id: I628f7b19e5b9abeb0fb08a573580dffa37f1b4e5
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Commit b0a2ee5567 ("drm/xe: prepare xe_gen_wa_oob to be multi-use")
introduced a call to basename(). The GNU version of this function is not
portable and fails to build with alternative libc implementations like
musl or bionic. This causes the following build error:
drivers/gpu/drm/xe/xe_gen_wa_oob.c:130:12: error: assignment to ‘const char *’ from ‘int’ makes pointer from integer without a cast [-Wint-conversion]
130 | fn = basename(fn);
| ^
While a POSIX version of basename() could be used, it would require a
separate header plus the behavior differs from GNU version in that it
might modify its argument. Not great.
Instead replace basename() with a strrchr() based implementation which
provides the same functionality and avoid portability issues.
Fixes: b0a2ee5567 ("drm/xe: prepare xe_gen_wa_oob to be multi-use")
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/all/20250820201612.2549797-1-cmllamas@google.com/
Change-Id: Icb43fd8af1fe25dd8828d3c6d20ec96d8768932f
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Steps on the way to v6.17-rc1
Resolve merge conflicts in:
drivers/dma-buf/heaps/system_heap.c
Cc: John Stultz <jstultz@google.com>
Change-Id: I091bc3daf3fefae11287bc79c80351f882ae51be
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Fix the following kernelci build error on non-CONFIG_SMP systems:
ERROR: modpost: vmlinux: local symbol 'set_task_cpu' was exported
Fixes: a085424328 ("ANDROID: Export some scheduler APIs for vendor modules")
Change-Id: Iba0c7f1c62fe0b2e4c386ad9adfa0ded61fd3733
Signed-off-by: Tiffany Yang <ynaffit@google.com>
This reverts commit b8ed6629a3.
Resolves trivial merge conflicts in:
drivers/misc/mei/Kconfig
drivers/misc/mei/hdcp/Kconfig
drivers/misc/mei/pxp/Kconfig
The lack of this commit causes an arm_allmodconfig build failure as
INTEL_MEI_ME is only expected to be selected for X86.
WARNING: unmet direct dependencies detected for INTEL_MEI_ME
Depends on [n]: X86 && PCI [=y]
Selected by [m]:
- INTEL_MEI_HDCP [=m] && (DRM_I915 [=n] || DRM_XE [=m])
- INTEL_MEI_PXP [=m] && (DRM_I915 [=n] || DRM_XE [=m])
drivers/gpu/drm/xe/xe_gen_wa_oob.c:130:5:
error: incompatible integer to pointer conversion assigning to
'const char *' from 'int' [-Wint-conversion]
130 | fn = basename(fn);
| ^ ~~~~~~~~~~~~
Change-Id: Ie9e307a4ae9ca88d5bcc4834da1bc1f7608d9699
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Commit cac5cefbad ("sched/smp: Make SMP unconditional") removed if
guards that used to be present around __migrate_task. We expose this
function to Android vendors, so KCI is giving the following build errors:
kernel/sched/core.c:2597:12: error: no previous prototype for ‘__migrate_task’ [-Werror=missing-prototypes]
2597 | struct rq *__migrate_task(struct rq *rq, struct rq_flags *rf,
| ^~~~~~~~~~~~~~
Remove the CONFIG_SMP if guards around the function prototype.
Change-Id: I849517a2cfe8a6883c6844fa77a905eb1a86c0f0
Fixes: 60d2746309 ("ANDROID: sched: allow access to critical common code for CPU Pause")
Signed-off-by: Tiffany Yang <ynaffit@google.com>
This reverts commit 8f15907b74.
The logic for ipi_desc was reworked upstream in commit ba1004f861
("arm64: smp: Support non-SGIs for IPIs"), so these functions are no
longer applicable. An updated version of this patch can be brought back
later if really needed.
Bug: 177393446
Fixes: ba1004f861 ("arm64: smp: Support non-SGIs for IPIs")
Change-Id: I0023cd9a428e24b67b752a86faa5dc3cc5564111
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Steps on the way to v6.17-rc1
Resolves merge conflicts in:
kernel/trace/power-traces.c
Change-Id: I3a3712bf41135f867910dace23bcb2447e467a71
Signed-off-by: Carlos Llamas <cmllamas@google.com>
This reverts commit 658dc70320.
Upstream hid 'trace_sched_swap_numa()' behind CONFIG_NUMA_BALANCING in
commit c2dbaf0af0 ("tracing: sched: Hide numa events under
CONFIG_NUMA_BALANCING"). This breaks ACK builds since migrate_swap()
uses such tracepoint and the CONFIG_NUMA_BALANCING build guards were
removed in https://r.android.com/3730131. This logic needs to be
reworked again if migrate_swap() is needed.
Bug: 438787298
Cc: Sai Harshini Nimmala <quic_snimmala@quicinc.com>
Change-Id: I08f62c35dbb81da664bc9c514722fd54bd07c12a
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Carlos Llamas
Amit Pundir
Alice Ryhl
Mike Yu
Isaac J. Manjarres
Frederick Mayle
Wake Liu
HONG Yifan
Alistair Delva
David Stevens
linjunjie1
Tiffany Yang
Bjorn Andersson