The "mwait_idle_with_hints" one was already there, some compiler
artifact now adds this ".constprop.0" suffix, cover that one too.
At some point we need to put these in a special bucket and show it
somewhere on the screen.
Noticed building the kernel on a fedora:32 system using:
gcc version 10.2.1 20200723 (Red Hat 10.2.1-1) (GCC)
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When I execute 'perf top' without HAVE_LIBBPF_SUPPORT, there exists the
following segmentation fault, skip the side-band event setup to fix it,
this is similar with commit 1101c872c8 ("perf record: Skip side-band
event setup if HAVE_LIBBPF_SUPPORT is not set").
[yangtiezhu@linux perf]$ ./perf top
<SNIP>
perf: Segmentation fault
Obtained 6 stack frames.
./perf(sighandler_dump_stack+0x5c) [0x12011b604]
[0xffffffc010]
./perf(perf_mmap__read_init+0x3e) [0x1201feeae]
./perf() [0x1200d715c]
/lib64/libpthread.so.0(+0xab9c) [0xffee10ab9c]
/lib64/libc.so.6(+0x128f4c) [0xffedc08f4c]
Segmentation fault
[yangtiezhu@linux perf]$
I use git bisect to find commit b38d85ef49 ("perf bpf: Decouple
creating the evlist from adding the SB event") is the first bad commit,
so also add the Fixes tag.
Committer testing:
First build perf explicitely disabling libbpf:
$ make NO_LIBBPF=1 O=/tmp/build/perf -C tools/perf install-bin && perf test python
Now make sure it isn't linked:
$ perf -vv | grep -w bpf
bpf: [ OFF ] # HAVE_LIBBPF_SUPPORT
$
$ nm ~/bin/perf | grep libbpf
$
And now try to run 'perf top':
# perf top
perf: Segmentation fault
-------- backtrace --------
perf[0x5bcd6d]
/lib64/libc.so.6(+0x3ca6f)[0x7fd0f5a66a6f]
perf(perf_mmap__read_init+0x1e)[0x5e1afe]
perf[0x4cc468]
/lib64/libpthread.so.0(+0x9431)[0x7fd0f645a431]
/lib64/libc.so.6(clone+0x42)[0x7fd0f5b2b912]
#
Applying this patch fixes the issue.
Fixes: b38d85ef49 ("perf bpf: Decouple creating the evlist from adding the SB event")
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Xuefeng Li <lixuefeng@loongson.cn>
Link: http://lore.kernel.org/lkml/1597753837-16222-1-git-send-email-yangtiezhu@loongson.cn
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
BPF basic filtering test fails on s390x (when vmlinux debuginfo is
utilized instead of /proc/kallsyms)
Info:
- bpf_probe_load installs the bpf code at do_epoll_wait.
- For s390x, do_epoll_wait resolves to 3 functions including inlines.
found inline addr: 0x43769e
Probe point found: __s390_sys_epoll_wait+6
found inline addr: 0x437290
Probe point found: do_epoll_wait+0
found inline addr: 0x4375d6
Probe point found: __se_sys_epoll_wait+6
- add_bpf_event creates evsel for every probe in a BPF object. This
results in 3 evsels.
Solution:
- Expected result = 50% of the samples to be collected from epoll_wait *
number of entries present in the evlist.
Committer testing:
# perf test 42
42: BPF filter :
42.1: Basic BPF filtering : Ok
42.2: BPF pinning : Ok
42.3: BPF prologue generation : Ok
42.4: BPF relocation checker : Ok
#
Signed-off-by: Sumanth Korikkar <sumanthk@linux.ibm.com>
Reviewed-by: Thomas Richter <tmricht@linux.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: bpf@vger.kernel.org
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Sven Schnelle <svens@linux.ibm.com>
LPU-Reference: 20200817072754.58344-1-sumanthk@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
can_nocow_extent and btrfs_cross_ref_exist both rely on a heuristic for
detecting a must cow condition which is not exactly accurate, but saves
unnecessary tree traversal. The incorrect assumption is that if the
extent was created in a generation smaller than the last snapshot
generation, it must be referenced by that snapshot. That is true, except
the snapshot could have since been deleted, without affecting the last
snapshot generation.
The original patch claimed a performance win from this check, but it
also leads to a bug where you are unable to use a swapfile if you ever
snapshotted the subvolume it's in. Make the check slower and more strict
for the swapon case, without modifying the general cow checks as a
compromise. Turning swap on does not seem to be a particularly
performance sensitive operation, so incurring a possibly unnecessary
btrfs_search_slot seems worthwhile for the added usability.
Note: Until the snapshot is competely cleaned after deletion,
check_committed_refs will still cause the logic to think that cow is
necessary, so the user must until 'btrfs subvolu sync' finished before
activating the swapfile swapon.
CC: stable@vger.kernel.org # 5.4+
Suggested-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
With my new locking code dbench is so much faster that I tripped over a
transaction abort from ENOSPC. This turned out to be because
btrfs_del_dir_entries_in_log was checking for ret == -ENOSPC, but this
function sets err on error, and returns err. So instead of properly
marking the inode as needing a full commit, we were returning -ENOSPC
and aborting in __btrfs_unlink_inode. Fix this by checking the proper
variable so that we return the correct thing in the case of ENOSPC.
The ENOENT needs to be checked, because btrfs_lookup_dir_item_index()
can return -ENOENT if the dir item isn't in the tree log (which would
happen if we hadn't fsync'ed this guy). We actually handle that case in
__btrfs_unlink_inode, so it's an expected error to get back.
Fixes: 4a500fd178 ("Btrfs: Metadata ENOSPC handling for tree log")
CC: stable@vger.kernel.org # 4.4+
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add note and comment about ENOENT ]
Signed-off-by: David Sterba <dsterba@suse.com>
For Intel controllers, SDHCI_RESET_ALL resets also CQHCI registers.
Normally, SDHCI_RESET_ALL is not used while CQHCI is enabled, but that can
happen on the error path. e.g. if mmc_cqe_recovery() fails, mmc_blk_reset()
is called which, for a eMMC that does not support HW Reset, will cycle the
bus power and the driver will perform SDHCI_RESET_ALL.
So whenever performing SDHCI_RESET_ALL ensure CQHCI is deactivated.
That will force the driver to reinitialize CQHCI when it is next used.
A similar change was done already for sdhci-msm, and other drivers using
CQHCI might benefit from a similar change, if they also have CQHCI reset
by SDHCI_RESET_ALL.
Fixes: 8ee82bda23 ("mmc: sdhci-pci: Add CQHCI support for Intel GLK")
Cc: stable@vger.kernel.org # 5.4.x: 0ffa6cfbd9: mmc: cqhci: Add cqhci_deactivate()
Cc: stable@vger.kernel.org # 5.4+
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Link: https://lore.kernel.org/r/20200819121848.16967-1-adrian.hunter@intel.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The AMD eMMC Controller can only use the tuned clock while in HS200 and
HS400 mode. If we switch to a different mode, we need to disable the
tuned clock. If we have previously performed tuning and switch back to
HS200 or HS400, we can re-enable the tuned clock.
Previously the tuned clock was not getting disabled when switching to
DDR52 which is part of the HS400 tuning sequence.
Fixes: 34597a3f60 ("mmc: sdhci-acpi: Add support for ACPI HID of AMD Controller with HS400")
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Link: https://lore.kernel.org/r/20200819125832.v2.1.Ie8f0689ec9f449203328b37409d1cf06b565f331@changeid
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The initialization done by bmips_cpu_setup() typically affects both
threads of a given core, on 7435 which supports 2 cores and 2 threads,
logical CPU number 2 and 3 would not run this initialization.
Fixes: 738a3f7902 ("MIPS: BMIPS: Add early CPU initialization code")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
When the BMIPS generic cpu-feature-overrides.h file was introduced,
cpu_has_inclusive_caches/MIPS_CPU_INCLUSIVE_CACHES was not set for
BMIPS5000 CPUs. Correct this when we have initialized the MIPS secondary
cache successfully.
Fixes: f337967d6d ("MIPS: BMIPS: Add cpu-feature-overrides.h")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
The mailbox CSR register has a write latency and requires a delay before
being read. This patch replaces readl_poll_timeout with read_poll_timeout
that allows to sleep before read.
The initial sleep was removed when the mailbox poll loop was replaced with
readl_poll_timeout.
Fixes: a79d471c65 ("crypto: qat - update timeout logic in put admin msg")
Signed-off-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
QString::sprintf() is deprecated in the latest Qt version, and spawns
a lot of warnings:
HOSTCXX scripts/kconfig/qconf.o
scripts/kconfig/qconf.cc: In member function ‘void ConfigInfoView::menuInfo()’:
scripts/kconfig/qconf.cc:1090:61: warning: ‘QString& QString::sprintf(const char*, ...)’ is deprecated: Use asprintf(), arg() or QTextStream instead [-Wdeprecated-declarations]
1090 | head += QString().sprintf("<a href=\"s%s\">", sym->name);
| ^
In file included from /usr/include/qt5/QtGui/qkeysequence.h:44,
from /usr/include/qt5/QtWidgets/qaction.h:44,
from /usr/include/qt5/QtWidgets/QAction:1,
from scripts/kconfig/qconf.cc:7:
/usr/include/qt5/QtCore/qstring.h:382:14: note: declared here
382 | QString &sprintf(const char *format, ...) Q_ATTRIBUTE_FORMAT_PRINTF(2, 3);
| ^~~~~~~
scripts/kconfig/qconf.cc:1099:60: warning: ‘QString& QString::sprintf(const char*, ...)’ is deprecated: Use asprintf(), arg() or QTextStream instead [-Wdeprecated-declarations]
1099 | head += QString().sprintf("<a href=\"s%s\">", sym->name);
| ^
In file included from /usr/include/qt5/QtGui/qkeysequence.h:44,
from /usr/include/qt5/QtWidgets/qaction.h:44,
from /usr/include/qt5/QtWidgets/QAction:1,
from scripts/kconfig/qconf.cc:7:
/usr/include/qt5/QtCore/qstring.h:382:14: note: declared here
382 | QString &sprintf(const char *format, ...) Q_ATTRIBUTE_FORMAT_PRINTF(2, 3);
| ^~~~~~~
scripts/kconfig/qconf.cc:1127:90: warning: ‘QString& QString::sprintf(const char*, ...)’ is deprecated: Use asprintf(), arg() or QTextStream instead [-Wdeprecated-declarations]
1127 | debug += QString().sprintf("defined at %s:%d<br><br>", _menu->file->name, _menu->lineno);
| ^
In file included from /usr/include/qt5/QtGui/qkeysequence.h:44,
from /usr/include/qt5/QtWidgets/qaction.h:44,
from /usr/include/qt5/QtWidgets/QAction:1,
from scripts/kconfig/qconf.cc:7:
/usr/include/qt5/QtCore/qstring.h:382:14: note: declared here
382 | QString &sprintf(const char *format, ...) Q_ATTRIBUTE_FORMAT_PRINTF(2, 3);
| ^~~~~~~
scripts/kconfig/qconf.cc: In member function ‘QString ConfigInfoView::debug_info(symbol*)’:
scripts/kconfig/qconf.cc:1150:68: warning: ‘QString& QString::sprintf(const char*, ...)’ is deprecated: Use asprintf(), arg() or QTextStream instead [-Wdeprecated-declarations]
1150 | debug += QString().sprintf("prompt: <a href=\"m%s\">", sym->name);
| ^
In file included from /usr/include/qt5/QtGui/qkeysequence.h:44,
from /usr/include/qt5/QtWidgets/qaction.h:44,
from /usr/include/qt5/QtWidgets/QAction:1,
from scripts/kconfig/qconf.cc:7:
/usr/include/qt5/QtCore/qstring.h:382:14: note: declared here
382 | QString &sprintf(const char *format, ...) Q_ATTRIBUTE_FORMAT_PRINTF(2, 3);
| ^~~~~~~
scripts/kconfig/qconf.cc: In static member function ‘static void ConfigInfoView::expr_print_help(void*, symbol*, const char*)’:
scripts/kconfig/qconf.cc:1225:59: warning: ‘QString& QString::sprintf(const char*, ...)’ is deprecated: Use asprintf(), arg() or QTextStream instead [-Wdeprecated-declarations]
1225 | *text += QString().sprintf("<a href=\"s%s\">", sym->name);
| ^
In file included from /usr/include/qt5/QtGui/qkeysequence.h:44,
from /usr/include/qt5/QtWidgets/qaction.h:44,
from /usr/include/qt5/QtWidgets/QAction:1,
from scripts/kconfig/qconf.cc:7:
/usr/include/qt5/QtCore/qstring.h:382:14: note: declared here
382 | QString &sprintf(const char *format, ...) Q_ATTRIBUTE_FORMAT_PRINTF(2, 3);
| ^~~~~~~
The documentation also says:
"Warning: We do not recommend using QString::asprintf() in new Qt code.
Instead, consider using QTextStream or arg(), both of which support
Unicode strings seamlessly and are type-safe."
Use QTextStream as suggested.
Reported-by: Robert Crawford <flacycads@cox.net>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
qconf is supposed to work with Qt4 and Qt5, but since commit
c4f7398bee ("kconfig: qconf: make debug links work again"),
building with Qt4 fails as follows:
HOSTCXX scripts/kconfig/qconf.o
scripts/kconfig/qconf.cc: In member function ‘void ConfigInfoView::clicked(const QUrl&)’:
scripts/kconfig/qconf.cc:1241:3: error: ‘qInfo’ was not declared in this scope; did you mean ‘setInfo’?
1241 | qInfo() << "Clicked link is empty";
| ^~~~~
| setInfo
scripts/kconfig/qconf.cc:1254:3: error: ‘qInfo’ was not declared in this scope; did you mean ‘setInfo’?
1254 | qInfo() << "Clicked symbol is invalid:" << data;
| ^~~~~
| setInfo
make[1]: *** [scripts/Makefile.host:129: scripts/kconfig/qconf.o] Error 1
make: *** [Makefile:606: xconfig] Error 2
qInfo() does not exist in Qt4. In my understanding, these call-sites
should be unreachable. Perhaps, qWarning(), assertion, or something
is better, but qInfo() is not the right one to use here, I think.
Fixes: c4f7398bee ("kconfig: qconf: make debug links work again")
Reported-by: Ronald Warsow <rwarsow@gmx.de>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
b->media->send_msg() requires rcu_read_lock(), as we can see
elsewhere in tipc, tipc_bearer_xmit, tipc_bearer_xmit_skb
and tipc_bearer_bc_xmit().
Syzbot has reported this issue as:
net/tipc/bearer.c:466 suspicious rcu_dereference_check() usage!
Workqueue: cryptd cryptd_queue_worker
Call Trace:
tipc_l2_send_msg+0x354/0x420 net/tipc/bearer.c:466
tipc_aead_encrypt_done+0x204/0x3a0 net/tipc/crypto.c:761
cryptd_aead_crypt+0xe8/0x1d0 crypto/cryptd.c:739
cryptd_queue_worker+0x118/0x1b0 crypto/cryptd.c:181
process_one_work+0x94c/0x1670 kernel/workqueue.c:2269
worker_thread+0x64c/0x1120 kernel/workqueue.c:2415
kthread+0x3b5/0x4a0 kernel/kthread.c:291
ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:293
So fix it by calling rcu_read_lock() in tipc_aead_encrypt_done()
for b->media->send_msg().
Fixes: fc1b6d6de2 ("tipc: introduce TIPC encryption & authentication")
Reported-by: syzbot+47bbc6b678d317cccbe0@syzkaller.appspotmail.com
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcf_ct_handle_fragments() shouldn't free the skb when ip_defrag() call
fails. Otherwise, we will cause a double-free bug.
In such cases, just return the error to the caller.
Fixes: b57dc7c13e ("net/sched: Introduce action ct")
Signed-off-by: Alaa Hleihel <alaa@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The number of output and input streams was never being reduced, eg when
processing received INIT or INIT_ACK chunks.
The effect is that DATA chunks can be sent with invalid stream ids
and then discarded by the remote system.
Fixes: 2075e50caf ("sctp: convert to genradix")
Signed-off-by: David Laight <david.laight@aculab.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- Remove pinctrl consumer properties, as they are handled by core
dt-schema,
- Document missing properties,
- Document missing PHY child node,
- Add "additionalProperties: false".
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Sergei Shtylyov <sergei.shtylyov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When receiving an IPv4 packet inside an IPv6 GRE packet, and the
IP6_TNL_F_RCV_DSCP_COPY flag is set on the tunnel, the IPv4 header would
get corrupted. This is due to the common ip6_tnl_rcv() function assuming
that the inner header is always IPv6. This patch checks the tunnel
protocol for IPv4 inner packets, but still defaults to IPv6.
Fixes: 308edfdf15 ("gre6: Cleanup GREv6 receive path, call common GRE functions")
Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
Signed-off-by: David S. Miller <davem@davemloft.net>
Set IP protocol mask only when IP protocol field is set.
This will allow flow-type ether with vlan rule which don't have
protocol field to apply.
ethtool -N ens5f4 flow-type ether proto 0x8100 vlan 0x600\
m 0x1FFF action 3 loc 16
Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
Haiyang Zhang says:
====================
hv_netvsc: Some fixes for the select_queue
This patch set includes two fixes for the select_queue process.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
netvsc_vf_xmit() / dev_queue_xmit() will call VF NIC’s ndo_select_queue
or netdev_pick_tx() again. They will use skb_get_rx_queue() to get the
queue number, so the “skb->queue_mapping - 1” will be used. This may
cause the last queue of VF not been used.
Use skb_record_rx_queue() here, so that the skb_get_rx_queue() called
later will get the correct queue number, and VF will be able to use
all queues.
Fixes: b3bf5666a5 ("hv_netvsc: defer queue selection to VF")
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When using vf_ops->ndo_select_queue, the number of queues of VF is
usually bigger than the synthetic NIC. This condition may happen
often.
Remove "unlikely" from the comparison of ndev->real_num_tx_queues.
Fixes: b3bf5666a5 ("hv_netvsc: defer queue selection to VF")
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In st21nfca_connectivity_event_received(), the return value of
devm_kzalloc() is unnecessarily cast from void*. Remove cast.
Issue identified with Coccinelle.
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In st_nci_hci_connectivity_event_received(), the return value of
devm_kzalloc() is unnecessarily cast from void*. Remove cast.
Issue identified with Coccinelle.
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In qed_rdma_destroy_cq() the result of dma_alloc_coherent() is cast from
void* unnecessarily. Remove cast.
Issue identified with Coccinelle.
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet says:
====================
tcp_mmap: optmizations
This series updates tcp_mmap reference tool to use best pratices.
First patch is using madvise(MADV_DONTNEED) to decrease pressure
on the socket lock.
Last patches try to use huge pages when available.
====================
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Arjun Roy <arjunroy@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
One down side of using TCP rx zerocopy is one extra TLB miss
per page after the mapping operation.
While if the application is using hugepages, the non zerocopy
recvmsg() will not have to pay these TLB costs.
This patch allows server side to use huge pages for
the non zero copy case, to allow fair comparisons when
both solutions use optimal conditions.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Arjun Roy <arjunroy@google.com>
Cc: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When TCP_ZEROCOPY_RECEIVE operation has been added,
I made the mistake of automatically un-mapping prior
content before mapping new pages.
This has the unfortunate effect of adding potentially long
MMU operations (like TLB flushes) while socket lock is held.
Using madvise(MADV_DONTNEED) right after pages has been used
has two benefits :
1) This releases pages sooner, allowing pages to be recycled
if they were part of a page pool in a NIC driver.
2) No more long unmap operations while preventing immediate
processing of incoming packets.
The cost of the added system call is small enough.
Arjun will submit a kernel patch allowing to opt out from
the unmap attempt in tcp_zerocopy_receive()
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Arjun Roy <arjunroy@google.com>
Cc: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, tcp sendmsg(MSG_ZEROCOPY) is building skbs with order-0 fragments.
Compared to standard sendmsg(), these skbs usually contain up to 16 fragments
on arches with 4KB page sizes, instead of two.
This adds considerable costs on various ndo_start_xmit() handlers,
especially when IOMMU is in the picture.
As high performance applications are often using huge pages,
we can try to combine adjacent pages belonging to same
compound page.
Tested on AMD Rome platform, with IOMMU, nominal single TCP flow speed
is roughly doubled (~55Gbit -> ~100Gbit), when user application
is using hugepages.
For reference, nominal single TCP flow speed on this platform
without MSG_ZEROCOPY is ~65Gbit.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simon Horman says:
====================
nfp: flower: add support for QinQ matching
Louis says:
Add new feature to the Netronome flower driver to enable QinQ offload.
This needed a bit of gymnastics in order to not break compatibility with
older firmware as the flow key sent to the firmware had to be updated
in order to make space for the extra field.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
When both the driver and the firmware supports QinQ the flow key
structure that is send to the firmware is updated as the old
method of matching on VLAN did not allow for space to add another
VLAN tag. VLAN flows can now also match on the tpid field, not
constrained to just 0x8100 as before.
Signed-off-by: Louis Peens <louis.peens@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The error path in libbpf.c:load_program() has calls to pr_warn()
which ends up for global_funcs tests to
test_global_funcs.c:libbpf_debug_print().
For the tests with no struct test_def::err_str initialized with a
string, it causes call of strstr() with NULL as the second argument
and it segfaults.
Fix it by calling strstr() only for non-NULL err_str.
Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20200820115843.39454-1-yauheni.kaliuta@redhat.com
7f0a838254 ("bpf, xdp: Maintain info on attached XDP BPF programs in net_device")
inadvertently changed which XDP mode is assumed when no mode flags are
specified explicitly. Previously, driver mode was preferred, if driver
supported it. If not, generic SKB mode was chosen. That commit changed default
to SKB mode always. This patch fixes the issue and restores the original
logic.
Fixes: 7f0a838254 ("bpf, xdp: Maintain info on attached XDP BPF programs in net_device")
Reported-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Tested-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://lore.kernel.org/bpf/20200820052841.1559757-1-andriin@fb.com
__smc_diag_dump() is potentially copying uninitialized kernel stack memory
into socket buffers, since the compiler may leave a 4-byte hole near the
beginning of `struct smcd_diag_dmbinfo`. Fix it by initializing `dinfo`
with memset().
Fixes: 4b1b7d3b30 ("net/smc: add SMC-D diag support")
Suggested-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Peilin Ye <yepeilin.cs@gmail.com>
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
IPv6 filters can occupy up to 4 slots and will exhaust HPFILTER
region much sooner. So, continue searching for free slots in the
HASH or NORMAL filter regions, as long as the rule's priority does
not conflict with existing rules in those regions.
Signed-off-by: Rahul Kundu <rahul.kundu@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
