Due to limitation of ABI checker, ABI differences can be reported if
symbol list of the reference ABI snapshot does not match. Add the
missing the other symbol lists.
The benefit lost by adding other symbol lists is that build_abi.sh would
not report missing symbols from abi_gki_aarch64_qcom symbol list since
symbol may be added by another contributor, but not into
abi_gki_aarch64_qcom list. This benefit can be re-acquired by pre-merge
checks.
Change-Id: Ie66bde15e7e8b085936faaecc06f0df91a42074d
Signed-off-by: Elliot Berman <eberman@codeaurora.org>
In QUPv3 common driver as part of geni_se_resources_init when bus
bandwidth voting is failed, we are dereferencing ctrl_dev which
can be NULL. Currently in spi-msm-geni driver we are initializing
ctrl_dev after calling geni_se_resource_init.
Fixed this by initializing ctrl_dev before calling geni_se_resource_init.
Change-Id: I68109c914e3f29eaac87f152c8732ea4f018502d
Signed-off-by: Anil Veshala <aveshala@codeaurora.org>
This change adds support to migrate from older fb_notifier
callback to a new notification framework defined by
panel_event_notifier driver and replaces file_operations with
proc_ops.
Change-Id: I71f5f5eccabc373c45aa848f2f2503bbb3e6a38c
Signed-off-by: BIVASH KUMAR SINGH <bkumarsi@codeaurora.org>
Add novatek chip id and firmware during boot up to
enable touch.
Change-Id: I382ae51c7bb4bf270c2e01b484281fcc24270062
Signed-off-by: Ravikanth Tuniki <rtunik@codeaurora.org>
Add CAPE & CAPE-V2 msm-id's to the list of soc-ids supported.
Change-Id: I740ad5a2504b2d2afcb16a59005b2af406d2c038
Signed-off-by: Auditya Bhattaram <audityab@codeaurora.org>
In QUPv3 common driver as part of geni_se_resources_init when bus
bandwidth voting is failed, we are dereferencing ctrl_dev which
can be NULL. Currently in msm_geni_serial driver we are initializing
ctrl_dev after calling geni_se_resource_init.
Fixed this by initializing ctrl_dev before calling geni_se_resource_init.
Change-Id: I744446d60e5b757e81c93cc1492d0ebc6cb507e2
Signed-off-by: Anvesh Salveru <asalveru@codeaurora.org>
When set_event() returns an error in setup_events(), the
error should only be returned for the case of -EPROBE_DEFER.
Set ret to 0 for all other errors to ensure a stale error
value isn't returned from setup_events().
Change-Id: I7a6a8e160e50f98c2a415341d39e8e9e71e61342
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Enable Neo pinctrl driver for the functioning of
Top Level Mode Multiplexer block (TLMM) on Neo platform.
Change-Id: I29fb4b519aa7ee14e0044ae9bb72725a240644e2
Signed-off-by: Amrit Anand <amrianan@codeaurora.org>
Make sure all switch statements can break to avoid fallthrough bugs.
Change-Id: I1a4bc821ea2fe77bf84377d1fae172343ef63742
Signed-off-by: Yue Ma <yuem@codeaurora.org>
* keystone/mirror-android12-5.10:
ANDROID: GKI: rockchip: Enable symbols for hid
ANDROID: GKI: rockchip: Enable symbols for rk808-regulator
UPSTREAM: arm64: kasan: don't populate vmalloc area for CONFIG_KASAN_VMALLOC
UPSTREAM: arm64: kasan: abstract _text and _end to KERNEL_START/END
UPSTREAM: arm64: Kconfig: support CONFIG_KASAN_VMALLOC
UPSTREAM: arm64: kaslr: support randomized module area with KASAN_VMALLOC
UPSTREAM: arm64: Kconfig: select KASAN_VMALLOC if KANSAN_GENERIC is enabled
FROMGIT: scs: Release kasan vmalloc poison in scs_free process
Signed-off-by: Daniel Norman <danielnorman@google.com>
Bug: 194515348
Bug: 201711661
Change-Id: I951e438e146a181adcef4a08e70c551a670ec78b
Before this patch, someone who wants to use VMAP_STACK when
KASAN_GENERIC enabled must explicitly select KASAN_VMALLOC.
>From Will's suggestion [1]:
> I would _really_ like to move to VMAP stack unconditionally, and
> that would effectively force KASAN_VMALLOC to be set if KASAN is in use
Because VMAP_STACK now depends on either HW_TAGS or KASAN_VMALLOC if
KASAN enabled, in order to make VMAP_STACK selected unconditionally,
we bind KANSAN_GENERIC and KASAN_VMALLOC together.
Note that SW_TAGS supports neither VMAP_STACK nor KASAN_VMALLOC now,
so this is the first step to make VMAP_STACK selected unconditionally.
Bind KANSAN_GENERIC and KASAN_VMALLOC together is supposed to cost more
memory at runtime, thus the alternative is using SW_TAGS KASAN instead.
[1]: https://lore.kernel.org/lkml/20210204150100.GE20815@willie-the-truck/
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Lecopzer Chen <lecopzer.chen@mediatek.com>
Link: https://lore.kernel.org/r/20210324040522.15548-6-lecopzer.chen@mediatek.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 201711661
(cherry picked from commit acc3042d62cb92c3776767f09f665511c903ef2d)
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
Change-Id: Ic7884e3acafa02361c8a250028ceebdb1780a49e
After KASAN_VMALLOC works in arm64, we can randomize module region
into vmalloc area now.
Test:
VMALLOC area ffffffc010000000 fffffffdf0000000
before the patch:
module_alloc_base/end ffffffc008b80000 ffffffc010000000
after the patch:
module_alloc_base/end ffffffdcf4bed000 ffffffc010000000
And the function that insmod some modules is fine.
Suggested-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Lecopzer Chen <lecopzer.chen@mediatek.com>
Link: https://lore.kernel.org/r/20210324040522.15548-5-lecopzer.chen@mediatek.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 201711661
(cherry picked from commit 31d02e7ab00873befd2cfb6e44581490d947c38b)
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
Change-Id: I513f205113752b396245e9e8d64b973fcd7ffcb1
Arm64 provides defined macro for KERNEL_START and KERNEL_END,
thus replace them by the abstration instead of using _text and _end.
Signed-off-by: Lecopzer Chen <lecopzer.chen@mediatek.com>
Acked-by: Andrey Konovalov <andreyknvl@gmail.com>
Tested-by: Andrey Konovalov <andreyknvl@gmail.com>
Tested-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20210324040522.15548-3-lecopzer.chen@mediatek.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 201711661
(cherry picked from commit 7d7b88ff5f8fd79a72baacc521407a3101f0ffae)
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
Change-Id: I806cb3d57e2064c272f3c87f0eb212ceea80c424
Linux support KAsan for VMALLOC since commit 3c5c3cfb9e
("kasan: support backing vmalloc space with real shadow memory")
Like how the MODULES_VADDR does now, just not to early populate
the VMALLOC_START between VMALLOC_END.
Before:
MODULE_VADDR: no mapping, no zero shadow at init
VMALLOC_VADDR: backed with zero shadow at init
After:
MODULE_VADDR: no mapping, no zero shadow at init
VMALLOC_VADDR: no mapping, no zero shadow at init
Thus the mapping will get allocated on demand by the core function
of KASAN_VMALLOC.
----------- vmalloc_shadow_start
| |
| |
| | <= non-mapping
| |
| |
|-----------|
|///////////|<- kimage shadow with page table mapping.
|-----------|
| |
| | <= non-mapping
| |
------------- vmalloc_shadow_end
|00000000000|
|00000000000| <= Zero shadow
|00000000000|
------------- KASAN_SHADOW_END
Signed-off-by: Lecopzer Chen <lecopzer.chen@mediatek.com>
Acked-by: Andrey Konovalov <andreyknvl@gmail.com>
Tested-by: Andrey Konovalov <andreyknvl@gmail.com>
Tested-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20210324040522.15548-2-lecopzer.chen@mediatek.com
[catalin.marinas@arm.com: add a build check on VMALLOC_START != MODULES_END]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 201711661
(cherry picked from commit 9a0732efa77418fc85b1bdc5ddee619e62f59545)
Signed-off-by: Yee Lee <yee.lee@mediatek.com>
Change-Id: I82c4b624685cc9e083d720904787df2326cb9cc0
* refs/heads/tmp-030acd3:
ANDROID: GKI: rockchip: Convert symbol to order by module
ANDROID: GKI: Add a symbol to symbol list
FROMLIST: arm64: mm: update max_pfn after memory hotplug
ANDROID: GKI: Update symbols to symbol list
UPSTREAM: erofs: fix 1 lcluster-sized pcluster for big pcluster
UPSTREAM: erofs: enable big pcluster feature
UPSTREAM: erofs: support decompress big pcluster for lz4 backend
UPSTREAM: erofs: support parsing big pcluster compact indexes
UPSTREAM: erofs: support parsing big pcluster compress indexes
UPSTREAM: erofs: adjust per-CPU buffers according to max_pclusterblks
UPSTREAM: erofs: add big physical cluster definition
UPSTREAM: erofs: fix up inplace I/O pointer for big pcluster
UPSTREAM: erofs: introduce physical cluster slab pools
UPSTREAM: erofs: introduce multipage per-CPU buffers
UPSTREAM: erofs: remove a void EROFS_VERSION macro set in Makefile
UPSTREAM: erofs: reserve physical_clusterbits[]
UPSTREAM: erofs: Clean up spelling mistakes found in fs/erofs
UPSTREAM: erofs: add on-disk compression configurations
UPSTREAM: erofs: introduce on-disk lz4 fs configurations
UPSTREAM: erofs: introduce erofs_sb_has_xxx() helpers
UPSTREAM: erofs: don't use erofs_map_blocks() any more
UPSTREAM: erofs: complete a missing case for inplace I/O
BACKPORT: UPSTREAM: mm: fs: invalidate bh_lrus for only cold path
FROMLIST: dma-buf: support users to change dma_buf.name
ANDROID: GKI: Update symbol list for vivo
BACKPORT: ASoC: soc-pcm: Get all BEs along DAPM path
Change-Id: I69fd05e026b660b07d9c68b0b2a526fe6aa5e57f
Signed-off-by: Sivasri Kumar, Vanka <svanka@codeaurora.org>
As part of log enhancements there were some unnecessary code changes
made and few parts of the code is deleted.
Adding back the deleted code and fixed few logical errors.
Change-Id: Ie4af0211347c110ffeaa6547b1304ae802b8bb0a
Signed-off-by: Anvesh Salveru <asalveru@codeaurora.org>
As a temporary workaround, reduce the USB buffer size to 64kB, which
translates to one TRB per request versus two TRBs if 128kB is used.
There is a chance that during high bandwidth use cases over QDSS, there
is a possible situation where a premature completion interrupt is
received when using chained transfers. This results to an SMMU fault,
as the USB controller is still accessing the TRB although we have
received the completion.
This should not affect overall QDSS bandwidth as the same number of
TRBs are allowed to be queued. The only difference is that it will
result in an interrupt for every transfer complete versus every other.
Change-Id: If45ba566768fdab6125adefba90ee883456b993a
Signed-off-by: Wesley Cheng <wcheng@codeaurora.org>
rpm_msgs address is calculated from memory allocated during write_batch.
Correct the offset from where rpm_msgs actually starts and assign req
pointer to point to correct address.
Also add list_del before freeing memory in rpmh_invalidate.
Change-Id: Id3ea4cb50f6def9556fc92cecaf0165af552c6f8
Signed-off-by: Maulik Shah <mkshah@codeaurora.org>
