-----BEGIN PGP SIGNATURE-----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=e9Bd
-----END PGP SIGNATURE-----
Merge tag 'locks-v5.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux
Pull file locking updates from Jeff Layton:
"Just a couple of small lease-related patches this cycle.
One from Ira to add a new tracepoint that fires during lease conflict
checks, and another patch from Amir to reduce false positives when
checking for lease conflicts"
* tag 'locks-v5.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux:
locks: eliminate false positive conflicts for write lease
locks: Add trace_leases_conflict
As Park Ju Hyung suggested:
"I'd like to suggest to write down an actual version of f2fs-tools
here as we've seen older versions of fsck doing even more damage
and the users might not have the latest f2fs-tools installed."
This patch give a more detailed info of how we fix such corruption
to user to avoid damageable repair with low version fsck.
Signed-off-by: Park Ju Hyung <qkrwngud825@gmail.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
blkoff_off might over 512 due to fs corrupt or security
vulnerability. That should be checked before being using.
Use ENTRIES_IN_SUM to protect invalid value in cur_data_blkoff.
Signed-off-by: Ocean Chen <oceanchen@google.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In umount, we give an constand time to handle pending discard, previously,
in __issue_discard_cmd() we missed to check timeout condition in loop,
result in delaying long time, fix it.
Signed-off-by: Heng Xiao <heng.xiao@unisoc.com>
[Chao Yu: add commit message]
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
The pinning of sensitive CR0 and CR4 bits caused a boot crash when loading
the kvm_intel module on a kernel compiled with CONFIG_PARAVIRT=n.
The reason is that the static key which controls the pinning is marked RO
after init. The kvm_intel module contains a CR4 write which requires to
update the static key entry list. That obviously does not work when the key
is in a RO section.
With CONFIG_PARAVIRT enabled this does not happen because the CR4 write
uses the paravirt indirection and the actual write function is built in.
As the key is intended to be immutable after init, move
native_write_cr0/4() out of line.
While at it consolidate the update of the cr4 shadow variable and store the
value right away when the pinning is initialized on a booting CPU. No point
in reading it back 20 instructions later. This allows to confine the static
key and the pinning variable to cpu/common and allows to mark them static.
Fixes: 8dbec27a24 ("x86/asm: Pin sensitive CR0 bits")
Fixes: 873d50d58f ("x86/asm: Pin sensitive CR4 bits")
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Xi Ruoyao <xry111@mengyan1223.wang>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Xi Ruoyao <xry111@mengyan1223.wang>
Acked-by: Kees Cook <keescook@chromium.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1907102140340.1758@nanos.tec.linutronix.de
On 32-bit x86 when building with clang-9, the 'division' loop gets turned
back into an inefficient division that causes a link error:
kernel/time/vsyscall.o: In function `update_vsyscall':
vsyscall.c:(.text+0xe3): undefined reference to `__udivdi3'
Use the existing __iter_div_u64_rem() function which is used to address the
same issue in other places.
Fixes: 44f57d788e ("timekeeping: Provide a generic update_vsyscall() implementation")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://lkml.kernel.org/r/20190710130206.1670830-1-arnd@arndb.de
We need to chain the earlier bios to the later ones, so that
submit_bio_wait waits on the bio that all the completions are
dispatched to.
Fixes: 6ad5b3255b ("xfs: use bios directly to read and write the log recovery buffers")
Reported-by: Dave Chinner <david@fromorbit.com>
Tested-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
clang points out that the computation of LOWMEM_PAGES causes a signed
integer overflow on 32-bit x86:
arch/x86/kernel/head32.c:83:20: error: signed shift result (0x100000000) requires 34 bits to represent, but 'int' only has 32 bits [-Werror,-Wshift-overflow]
(PAGE_TABLE_SIZE(LOWMEM_PAGES) << PAGE_SHIFT);
^~~~~~~~~~~~
arch/x86/include/asm/pgtable_32.h:109:27: note: expanded from macro 'LOWMEM_PAGES'
#define LOWMEM_PAGES ((((2<<31) - __PAGE_OFFSET) >> PAGE_SHIFT))
~^ ~~
arch/x86/include/asm/pgtable_32.h:98:34: note: expanded from macro 'PAGE_TABLE_SIZE'
#define PAGE_TABLE_SIZE(pages) ((pages) / PTRS_PER_PGD)
Use the _ULL() macro to make it a 64-bit constant.
Fixes: 1e620f9b23 ("x86/boot/32: Convert the 32-bit pgtable setup code from assembly to C")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20190710130522.1802800-1-arnd@arndb.de
When cross-compiling an out-of-tree build with an unclean source tree
directory, the build fails with:
/path/to/kernel/source/tree is not clean, please run 'make mrproper'
in the '/path/to/kernel/source/tree' directory.
However, doing so does not fix the problem, as "make mrproper" now
requires passing the target architecture to the make command, else it
won't remove $(srctree)/arch/$(SRCARCH)/include/generated.
"git ls-files -o" doesn't give a clue, as it doesn't list (empty)
directories, only files.
Improve usability by including the ARCH= option in the error output.
Fixes: a788b2ed81 ("kbuild: check arch/$(SRCARCH)/include/generated before out-of-tree build")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
A missing compression utility or other errors were not picked up by make
and an empty kernel image was produced. By removing the &&, errors will
no longer be ignored.
Signed-off-by: Harald Seiler <hws@denx.de>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
In old days, Kbuild always used an absolute path for $(srctree).
Since commit 890676c65d ("kbuild: Use relative path when building in
the source tree"), $(srctree) is '.' when O= was not passed from the
command line.
Yet, using absolute paths is useful in some cases even without O=, for
instance, to create a cscope file with absolute path tags.
'O=.' was known to work as a workaround to force Kbuild to use absolute
paths even when you are building in the source tree.
Since commit 25b146c5b8 ("kbuild: allow Kbuild to start from any
directory"), Kbuild is too clever to be tricked. Even if you pass 'O=.'
Kbuild notices you are building in the source tree, then use '.' for
$(srctree).
So, 'make O=. cscope' is no help to create absolute path tags.
We cannot force one or the other according to commit e93bc1a0ca
("Revert "kbuild: specify absolute paths for cscope""). Both of
relative path and absolute path have pros and cons.
This commit adds a new flag KBUILD_ABS_SRCTREE to allow users to
choose the absolute path for $(srctree).
'make KBUILD_ABS_SRCTREE=1 cscope' will work as a replacement of
'make O=. cscope'.
Reported-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Commit 25b146c5b8 ("kbuild: allow Kbuild to start from any directory")
deprecated KBUILD_SRCTREE.
It is only used in tools/testing/selftest/ to distinguish out-of-tree
build. Replace it with a new boolean flag, building_out_of_srctree.
I also replaced the conditional ($(srctree),.) because the next commit
will allow an absolute path to be used for $(srctree) even when building
in the source tree.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
We get a warning when build kernel W=1:
arch/x86/kvm/../../../virt/kvm/eventfd.c:48:1: warning: no previous prototype for ‘kvm_arch_irqfd_allowed’ [-Wmissing-prototypes]
kvm_arch_irqfd_allowed(struct kvm *kvm, struct kvm_irqfd *args)
^
The reason is kvm_arch_irqfd_allowed() is declared in arch/x86/kvm/irq.h,
which is not included by eventfd.c. Considering kvm_arch_irqfd_allowed()
is a weakly defined function in eventfd.c, remove the declaration to
kvm_host.h can fix this.
Signed-off-by: Yi Wang <wang.yi59@zte.com.cn>
Reviewed-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The field "page" is initialized to KVM_UNMAPPED_PAGE when it is not used
(i.e. when the memory lives outside kernel control). So this check will
always end up using kunmap even for memremap regions.
Fixes: e45adf665a ("KVM: Introduce a new guest mapping API")
Cc: stable@vger.kernel.org
Signed-off-by: KarimAllah Ahmed <karahmed@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
A greater or equal comparison on the unsigned int variable tmp_diff
is always true as unsigned ints are never negative. Hence the
comparison is redundant and can be removed.
Addresses-Coverity: ("Unsigned compared against 0")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Fix sparse warnings:
drivers/mmc/host/sdhci_am654.c:192:6: warning: symbol 'sdhci_j721e_4bit_set_clock' was not declared. Should it be static?
drivers/mmc/host/sdhci_am654.c:261:18: warning: symbol 'sdhci_j721e_8bit_ops' was not declared. Should it be static?
drivers/mmc/host/sdhci_am654.c:284:18: warning: symbol 'sdhci_j721e_4bit_ops' was not declared. Should it be static?
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
These days, the DMA mapping code must bounce buffers for any unsupported
address. If the driver needs to optimize for natively supported ranges,
then it should use dma_get_required_mask.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Marc Gonzalez <marc.w.gonzalez@free.fr>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Just like we do for all other block drivers. Especially as the limit
imposed at the moment might be way to pessimistic for iommus.
This also means we are not going to set a bounce limit for the queue, in
case we have a dma mask. On most architectures it was never needed, the
major hold out was x86-32 with PAE, but that has been fixed by now.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
As commit b6147490e6 ("mmc: tmio: split core functionality, DMA and
MFD glue") said, these MMC controllers use the IP from Panasonic.
TMIO (Toshiba Mobile IO) MMC was the first upstreamed user of this IP.
The common driver code was split and expanded as 'tmio-mmc-core', then
it became historical misnomer since 'tmio' is not the name of this IP.
In the discussion [1], we decide to keep this name as-is at least in
Linux driver level because renaming everything is a big churn.
However, DT should not be oriented to a particular project even though
it is mainly developed in Linux communities.
This is the misfortune only in Linux. Let's stop exporting it to other
projects, where there is no good reason to call this hardware "TMIO".
Rename the file to renesas,sdhi.txt. In fact, all the information in
this file is specific to the Renesas platform.
This commit also removes the first paragraph entirely. The DT-binding
should describe the hardware. It is strange to talk about Linux driver
internals such as how the drivers are probed, how platform data are
handed off, etc.
[1] https://www.spinics.net/lists/linux-mmc/msg46952.html
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
For Spreadtrum SD card voltage switching, besides regulator setting,
it also need switch related pin's state to output corresponding voltage.
This patch adds pin control operation to support voltage switch.
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
When changing SD card voltage signal for Spreadtrum SD host controller,
it also need to switch related pin's state. Thus add pinctrl properties'
description in documentation.
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
For Spreadtrum SD host controller, we can not use standard SD registers
to change and detect the I/O voltage signals, since our voltage regulator
for I/O is fixed in hardware, and no signals were connected to the SD
controller. Thus add Spreadtrum specific voltage switch ops to change
voltage instead of using standard SD host registers.
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Since the commit 133d624b1c ("dma: Introduce dma_max_mapping_size()")
provides a helper function to get the max mapping size, we can use
the function instead of the workaround code for swiotlb.
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The "powered_resume" in-parameter to mmc_sdio_init_card() has now become
redundant as all callers set it to 0. Therefore let's just drop it.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
The "powered_resume" in-parameter to mmc_sdio_reinit_card() has now become
redundant as all callers set it to 0. Therefore let's just drop it.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
It looks like the original idea behind always doing a re-initialization of
a removable SDIO card during system resume in mmc_sdio_resume(), is to try
to play safe to detect whether the card has been removed.
However, this seems like a really a bad idea as it will most likely screw
things up, especially when the card is expected to remain powered on during
system suspend by the SDIO func driver.
Let's fix this, simply by trusting that the detect work checks if the card
is alive and inserted, which is being scheduled at the PM_POST_SUSPEND
notification anyway.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
To use the so called powered-on re-initialization of an SDIO card, the
power to the card must obviously have stayed on. If not, the initialization
will simply fail.
In the runtime suspend case, the card is always powered off. Hence, let's
drop the support for powered-on re-initialization during runtime resume, as
it doesn't make sense.
Moreover, during a HW reset, the point is to cut the power to the card and
then do fresh re-initialization. Therefore drop the support for powered-on
re-initialization during HW reset.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Fixes: ca8971ca57 ("mmc: dw_mmc: Prevent runtime PM suspend when SDIO IRQs are enabled")
Tested-by: Douglas Anderson <dianders@chromium.org>
The comment in mmc_sdio_power_restore() belongs in mmc_sdio_reinit_card(),
which was created during a previous commit that re-factored some code. Fix
this by moving the comment into mmc_sdio_reinit_card().
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
The function mmc_sdio_power_restore() is called either from
mmc_sdio_runtime_resume() or from mmc_sdio_hw_reset(). Both callers either
claims/releases the host or require its callers to do so. Therefore let's
drop the redundant calls to mmc_claim|release_host() in
mmc_sdio_power_restore().
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Revert commit c522ad0637 ("ACPICA: Update table load object
initialization") as it causes systems to hang on attempts to load
OEM ACPI tables.
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Further cleanup from the SPDX fixup fallout for the recent ASPEED
series. aspeed_g4_defconfig, aspeed_g5_defconfig and multi_v5_defconfig
now compile. Smoke tested the g4 and g5 kernels under QEMU's
palmetto-bmc and romulus-bmc machines respectively.
Fixes: 35d8510ea3ad ("pinctrl: aspeed: Fix missed include")
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Link: https://lore.kernel.org/r/20190710032216.4088-1-andrew@aj.id.au
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Some SPDX churn made my fixes drop an important include
from the Aspeed pinctrl header. Fix it up.
Cc: Andrew Jeffery <andrew@aj.id.au>
Reported-by: Andrew Jeffery <andrew@aj.id.au>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
In some cases, using the 'truncate' command to extend a UDF file results
in a mismatch between the length of the file's extents (specifically, due
to incorrect length of the final NOT_ALLOCATED extent) and the information
(file) length. The discrepancy can prevent other operating systems
(i.e., Windows 10) from opening the file.
Two particular errors have been observed when extending a file:
1. The final extent is larger than it should be, having been rounded up
to a multiple of the block size.
B. The final extent is not shorter than it should be, due to not having
been updated when the file's information length was increased.
[JK: simplified udf_do_extend_final_block(), fixed up some types]
Fixes: 2c948b3f86 ("udf: Avoid IO in udf_clear_inode")
CC: stable@vger.kernel.org
Signed-off-by: Steven J. Magnani <steve@digidescorp.com>
Link: https://lore.kernel.org/r/1561948775-5878-1-git-send-email-steve@digidescorp.com
Signed-off-by: Jan Kara <jack@suse.cz>
Connections with legitimate tos values can get into usual connection
race. It can result in consumer reject. We don't want tos value or
protocol version to be demoted for such connections otherwise
piers would end up different tos values which can results in
no connection. Example a peer initiated connection with say
tos 8 while usual connection racing can get downgraded to tos 0
which is not desirable.
Patch fixes above issue introduced by commit
commit d021fabf52 ("rds: rdma: add consumer reject")
Reported-by: Yanjun Zhu <yanjun.zhu@oracle.com>
Tested-by: Yanjun Zhu <yanjun.zhu@oracle.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
The proper "tos" value needs to be returned
to user-space (sockopt RDS_INFO_CONNECTIONS).
Fixes: 3eb450367d ("rds: add type of service(tos) infrastructure")
Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
Reviewed-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Prior to
commit d021fabf52 ("rds: rdma: add consumer reject")
function "rds_rdma_cm_event_handler_cmn" would always honor a rejected
connection attempt by issuing a "rds_conn_drop".
The commit mentioned above added a "break", eliminating
the "fallthrough" case and made the "rds_conn_drop" rather conditional:
Now it only happens if a "consumer defined" reject (i.e. "rdma_reject")
carries an integer-value of "1" inside "private_data":
if (!conn)
break;
err = (int *)rdma_consumer_reject_data(cm_id, event, &len);
if (!err || (err && ((*err) == RDS_RDMA_REJ_INCOMPAT))) {
pr_warn("RDS/RDMA: conn <%pI6c, %pI6c> rejected, dropping connection\n",
&conn->c_laddr, &conn->c_faddr);
conn->c_proposed_version = RDS_PROTOCOL_COMPAT_VERSION;
rds_conn_drop(conn);
}
rdsdebug("Connection rejected: %s\n",
rdma_reject_msg(cm_id, event->status));
break;
/* FALLTHROUGH */
A number of issues are worth mentioning here:
#1) Previous versions of the RDS code simply rejected a connection
by calling "rdma_reject(cm_id, NULL, 0);"
So the value of the payload in "private_data" will not be "1",
but "0".
#2) Now the code has become dependent on host byte order and sizing.
If one peer is big-endian, the other is little-endian,
or there's a difference in sizeof(int) (e.g. ILP64 vs LP64),
the *err check does not work as intended.
#3) There is no check for "len" to see if the data behind *err is even valid.
Luckily, it appears that the "rdma_reject(cm_id, NULL, 0)" will always
carry 148 bytes of zeroized payload.
But that should probably not be relied upon here.
#4) With the added "break;",
we might as well drop the misleading "/* FALLTHROUGH */" comment.
This commit does _not_ address issue #2, as the sender would have to
agree on a byte order as well.
Here is the sequence of messages in this observed error-scenario:
Host-A is pre-QoS changes (excluding the commit mentioned above)
Host-B is post-QoS changes (including the commit mentioned above)
#1 Host-B
issues a connection request via function "rds_conn_path_transition"
connection state transitions to "RDS_CONN_CONNECTING"
#2 Host-A
rejects the incompatible connection request (from #1)
It does so by calling "rdma_reject(cm_id, NULL, 0);"
#3 Host-B
receives an "RDMA_CM_EVENT_REJECTED" event (from #2)
But since the code is changed in the way described above,
it won't drop the connection here, simply because "*err == 0".
#4 Host-A
issues a connection request
#5 Host-B
receives an "RDMA_CM_EVENT_CONNECT_REQUEST" event
and ends up calling "rds_ib_cm_handle_connect".
But since the state is already in "RDS_CONN_CONNECTING"
(as of #1) it will end up issuing a "rdma_reject" without
dropping the connection:
if (rds_conn_state(conn) == RDS_CONN_CONNECTING) {
/* Wait and see - our connect may still be succeeding */
rds_ib_stats_inc(s_ib_connect_raced);
}
goto out;
#6 Host-A
receives an "RDMA_CM_EVENT_REJECTED" event (from #5),
drops the connection and tries again (goto #4) until it gives up.
Tested-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
This reverts commit 5601245931.
RDS kept spinning inside function "rds_ib_post_reg_frmr", waiting for
"i_fastreg_wrs" to become incremented:
while (atomic_dec_return(&ibmr->ic->i_fastreg_wrs) <= 0) {
atomic_inc(&ibmr->ic->i_fastreg_wrs);
cpu_relax();
}
Looking at the original commit:
commit 5601245931 ("RDS: IB: split the mr registration and
invalidation path")
In there, the "rds_ib_mr_cqe_handler" was changed in the following
way:
void rds_ib_mr_cqe_handler(struct
rds_ib_connection *ic,
struct ib_wc *wc)
if (frmr->fr_inv) {
frmr->fr_state = FRMR_IS_FREE;
frmr->fr_inv = false;
atomic_inc(&ic->i_fastreg_wrs);
} else {
atomic_inc(&ic->i_fastunreg_wrs);
}
It looks like it's got it exactly backwards:
Function "rds_ib_post_reg_frmr" keeps track of the outstanding
requests via "i_fastreg_wrs".
Function "rds_ib_post_inv" keeps track of the outstanding requests
via "i_fastunreg_wrs" (post original commit). It also sets:
frmr->fr_inv = true;
However the completion handler "rds_ib_mr_cqe_handler" adjusts
"i_fastreg_wrs" when "fr_inv" had been true, and adjusts
"i_fastunreg_wrs" otherwise.
The original commit was done in the name of performance:
to remove the performance bottleneck
No performance benefit could be observed with a fixed-up version
of the original commit measured between two Oracle X7 servers,
both equipped with Mellanox Connect-X5 HCAs.
The prudent course of action is to revert this commit.
Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
RDS composite message(rdma + control) user notification needs to be
triggered once the full message is delivered and such a fix was
added as part of commit 941f8d55f6 ("RDS: RDMA: Fix the composite
message user notification"). But rds_send_remove_from_sock is missing
data part notify check and hence at times the user don't get
notification which isn't desirable.
One way is to fix the rds_send_remove_from_sock to check of that case
but considering the ordering complexity with completion handler and
rdma + control messages are always dispatched back to back in same send
context, just delaying the signaled completion on rmda work request also
gets the desired behaviour. i.e Notifying application only after
RDMA + control message send completes. So patch updates the earlier
fix with this approach. The delay signaling completions of rdma op
till the control message send completes fix was done by Venkat
Venkatsubra in downstream kernel.
Reviewed-and-tested-by: Zhu Yanjun <yanjun.zhu@oracle.com>
Reviewed-by: Gerd Rausch <gerd.rausch@oracle.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
clang warns:
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:251:2:
warning: variable 'rec_seq_sz' is used uninitialized whenever switch
default is taken [-Wsometimes-uninitialized]
default:
^~~~~~~
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:255:46: note:
uninitialized use occurs here
skip_static_post = !memcmp(rec_seq, &rn_be, rec_seq_sz);
^~~~~~~~~~
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c:239:16: note:
initialize the variable 'rec_seq_sz' to silence this warning
u16 rec_seq_sz;
^
= 0
1 warning generated.
This case statement was clearly designed to be one that should not be
hit during runtime because of the WARN_ON statement so just return early
to prevent copying uninitialized memory up into rn_be.
Fixes: d2ead1f360 ("net/mlx5e: Add kTLS TX HW offload support")
Link: https://github.com/ClangBuiltLinux/linux/issues/590
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Return value was changes to 'int' from void but this return statement
was not updated, or it slipped in via a merge.
Fixes: b5d9a834f4 ("net/tls: don't clear TX resync flag on error")
Signed-off-by: David S. Miller <davem@davemloft.net>
The OpenCAPI discovery and configuration specification has been
updated and introduces version 1.1 of the AFU descriptor template,
with new fields to better define the memory layout of an OpenCAPI
adapter.
The ocxl driver doesn't do much yet to support LPC memory but as we
start seeing (non-LPC) AFU images using the new template, this patch
updates the config space parsing code to avoid spitting a warning.
Signed-off-by: Alastair D'Silva <alastair@d-silva.org>
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Christophe Lombard <clombard@linux.vnet.ibm.com>
Acked-by: Andrew Donnellan <ajd@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20190605111545.19762-1-fbarrat@linux.ibm.com
