During a reply, the target gets woken up and then the priority of the
replier is restored. The order is such to allow the target to process
the reply ASAP. Otherwise, we risk the sender getting scheduled out
before the wakeup happens. This strategy reduces transaction latency.
However, a subsequent transaction from the same target could be started
before the priority of the replier gets restored. At this point we save
the wrong priority and it gets reinstated at the end of the transaction.
This patch allows the incoming transaction to detect the race condition
and save the correct next priority. Additionally, the replier will abort
its pending priority restore which allows the new transaction to always
run at the desired priority.
Bug: 148101660
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Change-Id: I6fec41ae1a1342023f78212ab1f984e26f068221
Refactor binder priority functions to take in 'struct binder_thread *'
instead of just 'struct task_struct *'. This allows access to other
thread fields used in subsequent patches. In any case, the same task
reference is still available under thread->task.
There is no functional impact from this patch.
Bug: 148101660
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Change-Id: I67b599884580d957d776500e467827e5035c99f6
Avoid making unnecessary stack copies of struct binder_priority and pass
the argument by reference instead. Rename 'desired_prio' to 'desired' to
match the usage in other priority functions.
There is no functional impact from this patch.
Bug: 148101660
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Change-Id: I66ff5305296e7b9dba56ed265236f2af518f66e0
The setup of node_prio is always the same, so just fold this logic into
binder_transaction_priority() to avoid duplication. Let's pass the node
reference instead, which also gives access to node->inherit_rt.
There is no functional impact from this patch.
Bug: 148101660
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Change-Id: Ib390204556e69c4bc8492cd9cd873773f9cdce42
db845c is not a mixed build yet, so need to add GKI
modules to it's module_outs for kleaf builds to
resolve hard failures in kleaf build for module copy.
Bug: 230519159
Test: tools/bazel run //common:db845c_dist
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
Change-Id: If3ce64a2b5f6b2f019a393f4674de30ac7437069
(cherry picked from commit eb38c6d799525c89bc3b1817dc22c34fa2281e09)
Enable zram and zsmalloc (dependency for zram) as
unprotected modules for aarch64. These are already
being used as modules by the vendor currently; so
needs to be unprotected.
Bug: 230519159
Test: TH
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
Change-Id: I7c617c1a24f6e083301cbed67d0d323388cbd622
(cherry picked from commit 0fe423433bb3fa0b5003409bbd59a972e6c235e2)
The default value of this option changed recently in
4f612ed3f7 ("kfence: default to dynamic branch instead of static keys
mode"). As there was no functional change here and no concern posed
about configurations relevant to Android, re-enable this option.
Bug: 234805737
Signed-off-by: Steve Muckle <smuckle@google.com>
Change-Id: I5b20a67225cd247fc54547083b3cbcde0135f1fb
This is a "weak" conversion which converts straight back to using pages
to keep the build working properly.
A full conversion should be performed at some point, hopefully by
someone familiar with the filesystem.
Fixes: 7e0a126519 ("mm,fs: Remove aops->readpage")
CC: Paul Lawrence <paullawrence@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: Id34db2c71743731d6a69b7b566a1f3a0a12dffe1
Steps on the way to 5.19-rc1
Resolves merge conflict in:
security/selinux/nlmsgtab.c
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I34db04ef9b75d4c781d0f6a42fa8d728e01ac0f4
Steps on the way to 5.19-rc1
Resolves merge conflicts with:
fs/crypto/inline_crypt.c
fs/crypto/keyring.c
fs/crypto/keysetup.c
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I624976a8b8d180b35e2f4e4ef282b07a44246682
Steps on the way to 5.19-rc1
Resolves merge conflicts in:
kernel/fork.c
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: Ie6468a4dcebacddea949d1ad5dfbaddea481e033
Steps on the way to 5.19-rc1
Resolves merge issues in:
include/linux/cpufreq.h
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I2f387e0728e85989b0f9853ec54bf196484e95d4
This reverts commit c6e5f9d7cf.
This causes merge problems with 5.19-rc1 and should no longer be needed
anymore.
If it is needed, please work to get the needed changes upstream first,
these should not be sticking around in the Android tree for no good
reason.
Bug: 169238689
Cc: Syed Rameez Mustafa <rameezmustafa@codeaurora.org>
Cc: Prasad Sodagudi <psodagud@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: Ifb7cf847cf185fb81b79b4fe31f8e5b8ccee8eba
This reverts commit 9da1f37103.
It causes merge problems with 5.19-rc1 and should no longer be needed at
all.
Bug: 169238689
Cc: Quentin Perret <qperret@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I27de2776905401b42c4e5cf2be784eae18f8816b
This reverts commit 1855071010.
This causes merge conflicts with 5.19-rc1 and should not be needed
anymore. Revert it for now and if it really is needed it can be added
back to a device-specific branch.
Bug: 178758017
Cc: Huang Yiwei <hyiwei@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: Id0bc0d42ae311f0af52a323fea5e58ed1b7e0bd7
There have been a lot of out-of-tree changes to the irq-gic-v3.c driver
that never got merged upstream. But other changes have been merged
upstream so resolving these merges has gotten way too complex.
So drop all out-of-tree changes for now to make merges easier. If the
out-of-tree changes really are needed, they can be added back later
(after they are at least submitted upstream first.)
Bug: 180471389
Bug: 190353898
Bug: 195914333
Bug: 168730134
Cc: Elliot Berman <eberman@codeaurora.org>
Cc: Maulik Shah <mkshah@codeaurora.org>
Cc: Neeraj Upadhyay <neeraju@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I97d8938fbe6c83911a4579695ff324ee4d83e7ef
Without initialization, it will be random data and hard for
vendor hook to decide.
Bug: 207739506
Change-Id: I278772d87eea38c03a40d4f0bef20ac8644e2ecd
Signed-off-by: Maria Yu <quic_aiquny@quicinc.com>
(cherry picked from commit 898e7ec950)
Building CONFIG_UAPI_HEADER_TEST=y with a Bionic (Android's libc) based
sysroot produces the following warning:
In file included from <built-in>:1:
./usr/include/linux/icmp.h:100:3: warning: declaration does not declare
anything [-Wmissing-declarations]
__be16 __unused;
^~~~~~
This is because Bionic defines __unused to expand to
__attribute__((__unused__)). Bionic pre-processes kernel headers and
redefines __unused to __linux_unused.
Do so here to avoid issues that only appear for Bionic based sysroot
UAPI header tests.
Link: 4ebdeebef7/libc/include/sys/cdefs.h (95)
Link: 4ebdeebef7/libc/kernel/tools/defaults.py (70)
Bug: 190019968
Bug: 234125788
Reported-by: Matthias Männich<maennich@google.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Change-Id: I2341953cbfce8e28b982c34df2df4b3b364d63a6
This reverts commit 99e629f14b ("PCI: dwc: Fix inverted condition of
DMA mask setup warning") and commit 660c486590 ("PCI: dwc: Set 32-bit
DMA mask for MSI target address allocation") in order to fix DW PCIe
host init issues on devices that support greater than 32-bit MSI target
address allocations. The bug was reported in [1] and the fix is
currently under review [2]. This change should be reverted once the
proper fix has merged upstream.
[1] https://lore.kernel.org/all/Yo0soniFborDl7+C@google.com/
[2] https://lore.kernel.org/all/20220525223316.388490-1-willmcvicker@google.com/
Bug: 232293973
Signed-off-by: Will McVicker <willmcvicker@google.com>
Change-Id: I2820743bbcbc09bfe4e12832b155e86997ef5cae
(cherry picked from commit e43c6c73302b4217e1df36d660c4f8af05b66551)
Pull devicetree updates from Rob Herring:
"Bindings:
- Convert smsc,lan91c111, qcom,spi-qup, qcom,msm-uartdm,
qcom,i2c-qup, qcom,gsbi, i2c-mt65xx, TI wkup_m3_ipc (and new
props), qcom,smp2p, TI timer, Mediatek gnss, Mediatek topckgen,
Mediatek apmixedsys, Mediatek infracfg, fsl,ls-extirq,
fsl,layerscape-dcfg, QCom PMIC SPMI, rda,8810pl-timer, Xilinx
zynqmp_ipi, uniphier-pcie, and Ilitek touchscreen controllers
- Convert various Arm Ltd peripheral IP bindings to schemas
- New bindings for Menlo board CPLD, DH electronics board CPLD,
Qualcomm Geni based QUP I2C, Renesas RZ/G2UL OSTM, Broafcom BCM4751
GNSS, MT6360 PMIC, ASIX USB Ethernet controllers, and
Microchip/SMSC LAN95xx USB Ethernet controllers
- Add vendor prefix for Enclustra
- Add various compatible string additions
- Various example fixes and cleanups
- Remove unused hisilicon,hi6220-reset binding
- Treewide fix properties missing type definition
- Drop some empty and unreferenced .txt bindings
- Documentation improvements for writing schemas
DT driver core:
- Drop static IRQ resources for DT platform devices as IRQ setup is
dynamic and drivers have all been converted to use
platform_get_irq() and friends
- Rework memory allocations and frees for overlays
- Continue overlay notifier callbacks on successful calls and add
unittests
- Handle 'interrupts-extended' in early DT IRQ setup
- Fix of_property_read_string() errors to match documentation
- Ignore disabled nodes in FDT API calls"
* tag 'devicetree-for-5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux: (86 commits)
of/irq: fix typo in comment
dt-bindings: Fix properties without any type
Revert "dt-bindings: mailbox: qcom-ipcc: add missing properties into example"
dt-bindings: input: touchscreen: ilitek_ts_i2c: Absorb ili2xxx bindings
dt-bindings: timer: samsung,exynos4210-mct: define strict clock order
dt-bindings: timer: samsung,exynos4210-mct: drop unneeded minItems
dt-bindings: timer: cdns,ttc: drop unneeded minItems
dt-bindings: mailbox: zynqmp_ipi: convert to yaml
dt-bindings: usb: ci-hdrc-usb2: fix node node for ethernet controller
dt-bindings: net: add schema for Microchip/SMSC LAN95xx USB Ethernet controllers
dt-bindings: net: add schema for ASIX USB Ethernet controllers
of/fdt: Ignore disabled memory nodes
dt-bindings: arm: fix typos in compatible
dt-bindings: mfd: Add bindings child nodes for the Mediatek MT6360
dt-bindings: display: convert Arm Komeda to DT schema
dt-bindings: display: convert Arm Mali-DP to DT schema
dt-bindings: display: convert Arm HDLCD to DT schema
dt-bindings: display: convert PL110/PL111 to DT schema
dt-bindings: arm: convert vexpress-config to DT schema
dt-bindings: arm: convert vexpress-sysregs to DT schema
...
Pull perf tool updates from Arnaldo Carvalho de Melo:
"Intel PT:
- Allow hardware tracing on KVM test programs. In this case, the VM
is not running an OS, but only the functions loaded into it by the
hypervisor test program, and conveniently, loaded at the same
virtual addresses.
- Improve documentation:
- Add link to perf wiki's page
- Cleanups:
- Delete now unused perf-with-kcore.sh script
- Remove unused machines__find_host()
ARM SPE (Statistical Profile Extensions):
- Add man page entry.
Vendor Events:
- Update various Intel event topics
- Update various microarch events
- Fix various cstate metrics
- Fix Alderlake metric groups
- Add sapphirerapids events
- Add JSON files for ARM Cortex A34, A35, A55, A510, A65, A73, A75,
A77, A78, A710, X1, X2 and Neoverse E1
- Update Cortex A57/A72
perf stat:
- Introduce stats for the user and system rusage times
perf c2c:
- Prep work to support ARM systems
perf annotate:
- Add --percent-limit option
perf lock:
- Add -t/--thread option for report
- Do not discard broken lock stats
perf bench:
- Add breakpoint benchmarks
perf test:
- Limit to only run executable scripts in tests
- Add basic perf record tests
- Add stat record+report test
- Add basic stat and topdown group test
- Skip several tests when the user hasn't permission to perform them
- Fix test case 81 ("perf record tests") on s390x
perf version:
- debuginfod support improvements
perf scripting python:
- Expose symbol offset and source information
perf build:
- Error for BPF skeletons without LIBBPF
- Use Python devtools for version autodetection rather than runtime
Miscellaneous:
- Add riscv64 support to 'perf jitdump'
- Various fixes/tidy ups related to cpu_map
- Fixes for handling Intel hybrid systems"
* tag 'perf-tools-for-v5.19-2022-05-23' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux: (122 commits)
perf intel-pt: Add guest_code support
perf kvm report: Add guest_code support
perf script: Add guest_code support
perf tools: Add guest_code support
perf tools: Factor out thread__set_guest_comm()
perf tools: Add machine to machines back pointer
perf vendors events arm64: Update Cortex A57/A72
perf vendors events arm64: Arm Neoverse E1
perf vendors events arm64: Arm Cortex-X2
perf vendors events arm64: Arm Cortex-X1
perf vendors events arm64: Arm Cortex-A710
perf vendors events arm64: Arm Cortex-A78
perf vendors events arm64: Arm Cortex-A77
perf vendors events arm64: Arm Cortex-A75
perf vendors events arm64: Arm Cortex-A73
perf vendors events arm64: Arm Cortex-A65
perf vendors events arm64: Arm Cortex-A510
perf vendors events arm64: Arm Cortex-A55
perf vendors events arm64: Arm Cortex-A35
perf vendors events arm64: Arm Cortex-A34
...
Pull checkpatch update from Gustavo Silva:
"kvmalloc() and kvzalloc() functions have now 2-factor multiplication
argument forms kvmalloc_array() and kvcalloc().
Add alloc-with-multiplies checks for these new functions"
* tag 'checkpatch-new-alloc-check-5.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gustavoars/linux:
checkpatch: add new alloc functions to alloc with multiplies check
