Eric Biggers [Sat, 15 Dec 2018 20:41:53 +0000 (12:41 -0800)]
crypto: skcipher - add might_sleep() to skcipher_walk_virt()
skcipher_walk_virt() can still sleep even with atomic=true, since that
only affects the later calls to skcipher_walk_done(). But,
skcipher_walk_virt() only has to allocate memory for some input data
layouts, so incorrectly calling it with preemption disabled can go
undetected. Use might_sleep() so that it's detected reliably.
Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Eric Biggers [Sat, 15 Dec 2018 20:40:17 +0000 (12:40 -0800)]
crypto: x86/chacha - avoid sleeping under kernel_fpu_begin()
Passing atomic=true to skcipher_walk_virt() only makes the later
skcipher_walk_done() calls use atomic memory allocations, not
skcipher_walk_virt() itself. Thus, we have to move it outside of the
preemption-disabled region (kernel_fpu_begin()/kernel_fpu_end()).
(skcipher_walk_virt() only allocates memory for certain layouts of the
input scatterlist, hence why I didn't notice this earlier...)
Reported-by: syzbot+9bf843c33f782d73ae7d@syzkaller.appspotmail.com Fixes: 4af78261870a ("crypto: x86/chacha20 - add XChaCha20 support") Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Fabio Estevam [Thu, 13 Dec 2018 09:52:32 +0000 (07:52 -0200)]
crypto: mxc-scc - fix build warnings on ARM64
The following build warnings are seen when building for ARM64 allmodconfig:
drivers/crypto/mxc-scc.c:181:20: warning: format '%d' expects argument of type 'int', but argument 5 has type 'size_t' {aka 'long unsigned int'} [-Wformat=]
drivers/crypto/mxc-scc.c:186:21: warning: format '%d' expects argument of type 'int', but argument 4 has type 'size_t' {aka 'long unsigned int'} [-Wformat=]
drivers/crypto/mxc-scc.c:277:21: warning: format '%d' expects argument of type 'int', but argument 4 has type 'size_t' {aka 'long unsigned int'} [-Wformat=]
drivers/crypto/mxc-scc.c:339:3: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
drivers/crypto/mxc-scc.c:340:3: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
Fix them by using the %zu specifier to print a size_t variable and using
a plain %x to print the result of a readl().
Signed-off-by: Fabio Estevam <festevam@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Corentin Labbe [Thu, 13 Dec 2018 08:36:37 +0000 (08:36 +0000)]
crypto: user - remove unused dump functions
This patch removes unused dump functions for crypto_user_stats.
There are remains of the copy/paste of crypto_user_base to
crypto_user_stat and I forgot to remove them.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
YueHaibing [Tue, 11 Dec 2018 08:11:59 +0000 (08:11 +0000)]
crypto: chelsio - remove set but not used variable 'kctx_len'
Fixes gcc '-Wunused-but-set-variable' warning:
drivers/crypto/chelsio/chcr_ipsec.c: In function 'chcr_ipsec_xmit':
drivers/crypto/chelsio/chcr_ipsec.c:674:33: warning:
variable 'kctx_len' set but not used [-Wunused-but-set-variable]
unsigned int flits = 0, ndesc, kctx_len;
It not used since commit 8362ea16f69f ("crypto: chcr - ESN for Inline IPSec Tx")
Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
crypto: ux500 - Use proper enum in hash_set_dma_transfer
Clang warns when one enumerated type is implicitly converted to another:
drivers/crypto/ux500/hash/hash_core.c:169:4: warning: implicit
conversion from enumeration type 'enum dma_data_direction' to different
enumeration type 'enum dma_transfer_direction' [-Wenum-conversion]
direction, DMA_CTRL_ACK | DMA_PREP_INTERRUPT);
^~~~~~~~~
1 warning generated.
dmaengine_prep_slave_sg expects an enum from dma_transfer_direction.
We know that the only direction supported by this function is
DMA_TO_DEVICE because of the check at the top of this function so we can
just use the equivalent value from dma_transfer_direction.
DMA_TO_DEVICE = DMA_MEM_TO_DEV = 1
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
crypto: ux500 - Use proper enum in cryp_set_dma_transfer
Clang warns when one enumerated type is implicitly converted to another:
drivers/crypto/ux500/cryp/cryp_core.c:559:5: warning: implicit
conversion from enumeration type 'enum dma_data_direction' to different
enumeration type 'enum dma_transfer_direction' [-Wenum-conversion]
direction, DMA_CTRL_ACK);
^~~~~~~~~
drivers/crypto/ux500/cryp/cryp_core.c:583:5: warning: implicit
conversion from enumeration type 'enum dma_data_direction' to different
enumeration type 'enum dma_transfer_direction' [-Wenum-conversion]
direction,
^~~~~~~~~
2 warnings generated.
dmaengine_prep_slave_sg expects an enum from dma_transfer_direction.
Because we know the value of the dma_data_direction enum from the
switch statement, we can just use the proper value from
dma_transfer_direction so there is no more conversion.
Dave Watson [Mon, 10 Dec 2018 19:59:59 +0000 (19:59 +0000)]
crypto: aesni - Add scatter/gather avx stubs, and use them in C
Add the appropriate scatter/gather stubs to the avx asm.
In the C code, we can now always use crypt_by_sg, since both
sse and asm code now support scatter/gather.
Introduce a new struct, aesni_gcm_tfm, that is initialized on
startup to point to either the SSE, AVX, or AVX2 versions of the
four necessary encryption/decryption routines.
GENX_OPTSIZE is still checked at the start of crypt_by_sg. The
total size of the data is checked, since the additional overhead
is in the init function, calculating additional HashKeys.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Introduce READ_PARTIAL_BLOCK macro, and use it in the two existing
partial block cases: AAD and the end of ENC_DEC. In particular,
the ENC_DEC case should be faster, since we read by 8/4 bytes if
possible.
This macro will also be used to read partial blocks between
enc_update and dec_update calls.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:59:11 +0000 (19:59 +0000)]
crypto: aesni - Move ghash_mul to GCM_COMPLETE
Prepare to handle partial blocks between scatter/gather calls.
For the last partial block, we only want to calculate the aadhash
in GCM_COMPLETE, and a new partial block macro will handle both
aadhash update and encrypting partial blocks between calls.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:58:56 +0000 (19:58 +0000)]
crypto: aesni - Fill in new context data structures
Fill in aadhash, aadlen, pblocklen, curcount with appropriate values.
pblocklen, aadhash, and pblockenckey are also updated at the end
of each scatter/gather operation, to be carried over to the next
operation.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:58:38 +0000 (19:58 +0000)]
crypto: aesni - Merge avx precompute functions
The precompute functions differ only by the sub-macros
they call, merge them to a single macro. Later diffs
add more code to fill in the gcm_context_data structure,
this allows changes in a single place.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:58:19 +0000 (19:58 +0000)]
crypto: aesni - Split AAD hash calculation to separate macro
AAD hash only needs to be calculated once for each scatter/gather operation.
Move it to its own macro, and call it from GCM_INIT instead of
INITIAL_BLOCKS.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:57:36 +0000 (19:57 +0000)]
crypto: aesni - support 256 byte keys in avx asm
Add support for 192/256-bit keys using the avx gcm/aes routines.
The sse routines were previously updated in e31ac32d3b (Add support
for 192 & 256 bit keys to AESNI RFC4106).
Instead of adding an additional loop in the hotpath as in e31ac32d3b,
this diff instead generates separate versions of the code using macros,
and the entry routines choose which version once. This results
in a 5% performance improvement vs. adding a loop to the hot path.
This is the same strategy chosen by the intel isa-l_crypto library.
The key size checks are removed from the c code where appropriate.
Note that this diff depends on using gcm_context_data - 256 bit keys
require 16 HashKeys + 15 expanded keys, which is larger than
struct crypto_aes_ctx, so they are stored in struct gcm_context_data.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dave Watson [Mon, 10 Dec 2018 19:56:45 +0000 (19:56 +0000)]
crypto: aesni - Merge GCM_ENC_DEC
The GCM_ENC_DEC routines for AVX and AVX2 are identical, except they
call separate sub-macros. Pass the macros as arguments, and merge them.
This facilitates additional refactoring, by requiring changes in only
one place.
The GCM_ENC_DEC macro was moved above the CONFIG_AS_AVX* ifdefs,
since it will be used by both AVX and AVX2.
Signed-off-by: Dave Watson <davejwatson@fb.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Fix this by sanitizing protocol before using it to index proto_tab.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
Willem de Bruijn [Sat, 22 Dec 2018 21:53:45 +0000 (16:53 -0500)]
packet: validate address length if non-zero
Validate packet socket address length if a length is given. Zero
length is equivalent to not setting an address.
Fixes: 99137b7888f4 ("packet: validate address length") Reported-by: Ido Schimmel <idosch@idosch.org> Signed-off-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Fix this by sanitizing proto before using it to index proto_tab.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
Fix this by sanitizing protocol before using it to index proto_tab.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
Fix this by sanitizing flen before using it to index filter at line 1101:
switch (filter[flen - 1].code) {
and through pc at line 1040:
const struct sock_filter *ftest = &filter[pc];
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
Linus Torvalds [Sat, 22 Dec 2018 23:03:00 +0000 (15:03 -0800)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"This is two simple target fixes and one discard related I/O starvation
problem in sd.
The discard problem occurs because the discard page doesn't have a
mempool backing so if the allocation fails due to memory pressure, we
then lose the forward progress we require if the writeout is on the
same device. The fix is to back it with a mempool"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: sd: use mempool for discard special page
scsi: target: iscsi: cxgbit: add missing spin_lock_init()
scsi: target: iscsi: cxgbit: fix csk leak
Linus Torvalds [Sat, 22 Dec 2018 22:29:21 +0000 (14:29 -0800)]
Merge tag 'compiler-attributes-for-linus-v4.20' of https://github.com/ojeda/linux
Pull compiler_types.h fix from Miguel Ojeda:
"A cleanup for userspace in compiler_types.h: don't pollute userspace
with macro definitions (Xiaozhou Liu)
This is harmless for the kernel, but v4.19 was released with a few
macros exposed to userspace as the patch explains; which this removes,
so it *could* happen that we break something for someone (although
leaving inline redefined is probably worse)"
* tag 'compiler-attributes-for-linus-v4.20' of https://github.com/ojeda/linux:
include/linux/compiler_types.h: don't pollute userspace with macro definitions
commit 55956b59df33 ("vfs: Allow userns root to call mknod on owned filesystems.")
enabled mknod() in user namespaces for userns root if CAP_MKNOD is
available. However, these device nodes are useless since any filesystem
mounted from a non-initial user namespace will set the SB_I_NODEV flag on
the filesystem. Now, when a device node s created in a non-initial user
namespace a call to open() on said device node will fail due to:
The problem with this is that as of the aforementioned commit mknod()
creates partially functional device nodes in non-initial user namespaces.
In particular, it has the consequence that as of the aforementioned commit
open() will be more privileged with respect to device nodes than mknod().
Before it was the other way around. Specifically, if mknod() succeeded
then it was transparent for any userspace application that a fatal error
must have occured when open() failed.
All of this breaks multiple userspace workloads and a widespread assumption
about how to handle mknod(). Basically, all container runtimes and systemd
live by the slogan "ask for forgiveness not permission" when running user
namespace workloads. For mknod() the assumption is that if the syscall
succeeds the device nodes are useable irrespective of whether it succeeds
in a non-initial user namespace or not. This logic was chosen explicitly
to allow for the glorious day when mknod() will actually be able to create
fully functional device nodes in user namespaces.
A specific problem people are already running into when running 4.18 rc
kernels are failing systemd services. For any distro that is run in a
container systemd services started with the PrivateDevices= property set
will fail to start since the device nodes in question cannot be
opened (cf. the arguments in [1]).
Full disclosure, Seth made the very sound argument that it is already
possible to end up with partially functional device nodes. Any filesystem
mounted with MS_NODEV set will allow mknod() to succeed but will not allow
open() to succeed. The difference to the case here is that the MS_NODEV
case is transparent to userspace since it is an explicitly set mount option
while the SB_I_NODEV case is an implicit property enforced by the kernel
and hence opaque to userspace.
x86/efi: Don't unmap EFI boot services code/data regions for EFI_OLD_MEMMAP and EFI_MIXED_MODE
The following commit:
d5052a7130a6 ("x86/efi: Unmap EFI boot services code/data regions from efi_pgd")
forgets to take two EFI modes into consideration, namely EFI_OLD_MEMMAP and
EFI_MIXED_MODE:
- EFI_OLD_MEMMAP is a legacy way of mapping EFI regions into swapper_pg_dir
using ioremap() and init_memory_mapping(). This feature can be enabled by
passing "efi=old_map" as kernel command line argument. But,
efi_unmap_pages() unmaps EFI boot services code/data regions *only* from
efi_pgd and hence cannot be used for unmapping EFI boot services code/data
regions from swapper_pg_dir.
Introduce a temporary fix to not unmap EFI boot services code/data regions
when EFI_OLD_MEMMAP is enabled while working on a real fix.
- EFI_MIXED_MODE is another feature where a 64-bit kernel runs on a
64-bit platform crippled by a 32-bit firmware. To support EFI_MIXED_MODE,
all RAM (i.e. namely EFI regions like EFI_CONVENTIONAL_MEMORY,
EFI_LOADER_<CODE/DATA>, EFI_BOOT_SERVICES_<CODE/DATA> and
EFI_RUNTIME_CODE/DATA regions) is mapped into efi_pgd all the time to
facilitate EFI runtime calls access it's arguments in 1:1 mode.
Hence, don't unmap EFI boot services code/data regions when booted in mixed mode.
Signed-off-by: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com> Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Bhupesh Sharma <bhsharma@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Rik van Riel <riel@surriel.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-efi@vger.kernel.org Link: http://lkml.kernel.org/r/20181222022234.7573-1-sai.praneeth.prakhya@intel.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
powerpc: Fix HMIs on big-endian with CONFIG_RELOCATABLE=y
HMIs will crash the kernel due to
BRANCH_LINK_TO_FAR(hmi_exception_realmode)
Calling into the OPD instead of the actual code.
Fixes: 2337d207288f ("powerpc/64: CONFIG_RELOCATABLE support for hmi interrupts") Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[mpe: Use DOTSYM() rather than #ifdef] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Rob Herring [Wed, 5 Dec 2018 19:50:28 +0000 (13:50 -0600)]
macintosh: Use of_node_name_{eq, prefix} for node name comparisons
Convert string compares of DT node names to use of_node_name_{eq,prefix}
helpers instead. This removes direct access to the node name pointer.
This changes a single case insensitive node name comparison to case
sensitive for "ata4". This is the only instance of a case insensitive
comparison for all the open coded node name comparisons on powerpc.
Searching the commit history, there doesn't appear to be any reason for
it to be case insensitive.
A couple of open coded iterating thru the child node names are converted
to use for_each_child_of_node() instead.
Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Rob Herring [Wed, 5 Dec 2018 19:50:25 +0000 (13:50 -0600)]
ide: Use of_node_name_eq for node name comparisons
Convert string compares of DT node names to use of_node_name_eq helper
instead. This removes direct access to the node name pointer.
Cc: "David S. Miller" <davem@davemloft.net> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: linux-ide@vger.kernel.org Cc: linuxppc-dev@lists.ozlabs.org Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Rob Herring [Wed, 5 Dec 2018 19:50:17 +0000 (13:50 -0600)]
powerpc/pseries/pmem: Convert to %pOFn instead of device_node.name
In preparation to remove the node name pointer from struct
device_node, convert printf users to use the %pOFn format specifier.
pmem.c was recently added and missed the initial conversion.
Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Michael Ellerman [Tue, 27 Nov 2018 08:18:05 +0000 (19:18 +1100)]
powerpc/mm: Remove very old comment in hash-4k.h
This comment talks about PTEs being 64-bits and PMD/PGD being 32-bits,
but that hasn't been true since 2005 when David Gibson implemented
4-level page tables in the commit titled "Four level pagetables for
ppc64".
Remove it.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Michael Ellerman [Tue, 27 Nov 2018 08:16:44 +0000 (19:16 +1100)]
powerpc/pseries: Fix node leak in update_lmb_associativity_index()
In update_lmb_associativity_index() we lookup dr_node using
of_find_node_by_path() which takes a reference for us. In the
non-error case we forget to drop the reference. Note that
find_aa_index() does modify properties of the node, but doesn't need
an extra reference held once it's returned.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Scott Wood [Sat, 22 Dec 2018 03:32:51 +0000 (21:32 -0600)]
powerpc/dts/fsl: Fix dtc-flagged interrupt errors
mpc8641_hpcn was updated to 4-cell interrupt specifiers, but
PCI interrupt-map was not updated. It was also missing #interrupt-cells
on the outer PCI buses.
p1020rdb-pc was updated to 4-cell interrupt specifiers, but
the ethernet-phy nodes weren't updated.
mpc832x_rdb had an invalid "interrupts = <0>" on the ethernet-phy nodes.
Besides being the wrong number of cells, 0 is not a valid IPIC interrupt
according to ipic.c. Presumably it was meant to indicate that these
PHYs are not connected to an interrupt.
Scott Wood [Wed, 31 Oct 2018 06:57:35 +0000 (14:57 +0800)]
powerpc/fsl: Use new clockgen binding
The driver retains compatibility with old device trees, but we don't
want the old nodes lying around to be copied, or used as a reference
(some of the mux options are incorrect), or even just being clutter.
Signed-off-by: Scott Wood <oss@buserror.net> Signed-off-by: Tang Yuantian <andy.tang@nxp.com>
[scottwood: removed sysclk node added by Andy] Signed-off-by: Scott Wood <oss@buserror.net>
Christophe Leroy [Mon, 10 Dec 2018 11:41:29 +0000 (11:41 +0000)]
powerpc/83xx: handle machine check caused by watchdog timer
When the watchdog timer is set in interrupt mode, it causes a
machine check when it times out. The purpose of this mode is to
ease debugging, not to crash the kernel and reboot the machine.
This patch implements a special handling for that, in order to not
crash the kernel if the watchdog times out while in interrupt or
within the idle task.
Omar Sandoval [Sat, 22 Dec 2018 02:45:18 +0000 (18:45 -0800)]
xfs: reallocate realtime summary cache on growfs
At mount time, we allocate m_rsum_cache with the number of realtime
bitmap blocks. However, xfs_growfs_rt() can increase the number of
realtime bitmap blocks. Using the cache after this happens may access
out of the bounds of the cache. Fix it by reallocating the cache in this
case.
Fixes: 355e3532132b ("xfs: cache minimum realtime summary level") Signed-off-by: Omar Sandoval <osandov@fb.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
swiotlb will only bounce buffer when the effective dma address for the
device is smaller than the actual DMA range. Instead of flipping between
the swiotlb and nommu ops for FSL SOCs that have the second outbound
window just don't set the bus dma_mask in this case.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Scott Wood <oss@buserror.net>
Linus Torvalds [Fri, 21 Dec 2018 22:59:00 +0000 (14:59 -0800)]
Merge branch 'akpm' (patches from Andrew)
Merge misc fixes from Andrew Morton:
"4 fixes"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
mm, page_alloc: fix has_unmovable_pages for HugePages
fork,memcg: fix crash in free_thread_stack on memcg charge fail
mm: thp: fix flags for pmd migration when split
mm, memory_hotplug: initialize struct pages for the full memory section
The reason is we do not pass the Head to page_hstate(), and so, the call
to compound_order() in page_hstate() returns 0, so we end up checking
all hstates's size to match PAGE_SIZE.
Obviously, we do not find any hstate matching that size, and we return
NULL. Then, we dereference that NULL pointer in
hugepage_migration_supported() and we got the #PF from above.
Fix that by getting the head page before calling page_hstate().
Also, since gigantic pages span several pageblocks, re-adjust the logic
for skipping pages. While are it, we can also get rid of the
round_up().
[osalvador@suse.de: remove round_up(), adjust skip pages logic per Michal] Link: http://lkml.kernel.org/r/20181221062809.31771-1-osalvador@suse.de Link: http://lkml.kernel.org/r/20181217225113.17864-1-osalvador@suse.de Signed-off-by: Oscar Salvador <osalvador@suse.de> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: David Hildenbrand <david@redhat.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Pavel Tatashin <pavel.tatashin@microsoft.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Rik van Riel [Fri, 21 Dec 2018 22:30:54 +0000 (14:30 -0800)]
fork,memcg: fix crash in free_thread_stack on memcg charge fail
Commit 9b6f7e163cd0 ("mm: rework memcg kernel stack accounting") will
result in fork failing if allocating a kernel stack for a task in
dup_task_struct exceeds the kernel memory allowance for that cgroup.
Unfortunately, it also results in a crash.
This is due to the code jumping to free_stack and calling
free_thread_stack when the memcg kernel stack charge fails, but without
tsk->stack pointing at the freshly allocated stack.
This in turn results in the vfree_atomic in free_thread_stack oopsing
with a backtrace like this:
Peter Xu [Fri, 21 Dec 2018 22:30:50 +0000 (14:30 -0800)]
mm: thp: fix flags for pmd migration when split
When splitting a huge migrating PMD, we'll transfer all the existing PMD
bits and apply them again onto the small PTEs. However we are fetching
the bits unconditionally via pmd_soft_dirty(), pmd_write() or
pmd_yound() while actually they don't make sense at all when it's a
migration entry. Fix them up. Since at it, drop the ifdef together as
not needed.
Note that if my understanding is correct about the problem then if
without the patch there is chance to lose some of the dirty bits in the
migrating pmd pages (on x86_64 we're fetching bit 11 which is part of
swap offset instead of bit 2) and it could potentially corrupt the
memory of an userspace program which depends on the dirty bit.
Link: http://lkml.kernel.org/r/20181213051510.20306-1-peterx@redhat.com Signed-off-by: Peter Xu <peterx@redhat.com> Reviewed-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Reviewed-by: William Kucharski <william.kucharski@oracle.com> Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Dave Jiang <dave.jiang@intel.com> Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> Cc: Souptick Joarder <jrdr.linux@gmail.com> Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Zi Yan <zi.yan@cs.rutgers.edu> Cc: <stable@vger.kernel.org> [4.14+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Mikhail Zaslonko [Fri, 21 Dec 2018 22:30:46 +0000 (14:30 -0800)]
mm, memory_hotplug: initialize struct pages for the full memory section
If memory end is not aligned with the sparse memory section boundary,
the mapping of such a section is only partly initialized. This may lead
to VM_BUG_ON due to uninitialized struct page access from
is_mem_section_removable() or test_pages_in_a_zone() function triggered
by memory_hotplug sysfs handlers:
Here are the the panic examples:
CONFIG_DEBUG_VM=y
CONFIG_DEBUG_VM_PGFLAGS=y
kernel parameter mem=2050M
--------------------------
page:000003d082008000 is uninitialized and poisoned
page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p))
Call Trace:
( test_pages_in_a_zone+0xde/0x160)
show_valid_zones+0x5c/0x190
dev_attr_show+0x34/0x70
sysfs_kf_seq_show+0xc8/0x148
seq_read+0x204/0x480
__vfs_read+0x32/0x178
vfs_read+0x82/0x138
ksys_read+0x5a/0xb0
system_call+0xdc/0x2d8
Last Breaking-Event-Address:
test_pages_in_a_zone+0xde/0x160
Kernel panic - not syncing: Fatal exception: panic_on_oops
kernel parameter mem=3075M
--------------------------
page:000003d08300c000 is uninitialized and poisoned
page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p))
Call Trace:
( is_mem_section_removable+0xb4/0x190)
show_mem_removable+0x9a/0xd8
dev_attr_show+0x34/0x70
sysfs_kf_seq_show+0xc8/0x148
seq_read+0x204/0x480
__vfs_read+0x32/0x178
vfs_read+0x82/0x138
ksys_read+0x5a/0xb0
system_call+0xdc/0x2d8
Last Breaking-Event-Address:
is_mem_section_removable+0xb4/0x190
Kernel panic - not syncing: Fatal exception: panic_on_oops
Fix the problem by initializing the last memory section of each zone in
memmap_init_zone() till the very end, even if it goes beyond the zone end.
Michal said:
: This has alwways been problem AFAIU. It just went unnoticed because we
: have zeroed memmaps during allocation before f7f99100d8d9 ("mm: stop
: zeroing memory during allocation in vmemmap") and so the above test
: would simply skip these ranges as belonging to zone 0 or provided a
: garbage.
:
: So I guess we do care for post f7f99100d8d9 kernels mostly and
: therefore Fixes: f7f99100d8d9 ("mm: stop zeroing memory during
: allocation in vmemmap")
Link: http://lkml.kernel.org/r/20181212172712.34019-2-zaslonko@linux.ibm.com Fixes: f7f99100d8d9 ("mm: stop zeroing memory during allocation in vmemmap") Signed-off-by: Mikhail Zaslonko <zaslonko@linux.ibm.com> Reviewed-by: Gerald Schaefer <gerald.schaefer@de.ibm.com> Suggested-by: Michal Hocko <mhocko@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: Pasha Tatashin <Pavel.Tatashin@microsoft.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull sparc fixes from David Miller:
"Just some small fixes here and there, and a refcount leak in a serial
driver, nothing serious"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
serial/sunsu: fix refcount leak
sparc: Set "ARCH: sunxx" information on the same line
sparc: vdso: Drop implicit common-page-size linker flag
Pull more networking fixes from David Miller:
"Some more bug fixes have trickled in, we have:
1) Local MAC entries properly in mscc driver, from Allan W. Nielsen.
2) Eric Dumazet found some more of the typical "pskb_may_pull() -->
oops forgot to reload the header pointer" bugs in ipv6 tunnel
handling.
3) Bad SKB socket pointer in ipv6 fragmentation handling, from Herbert
Xu.
4) Overflow fix in sk_msg_clone(), from Vakul Garg.
5) Validate address lengths in AF_PACKET, from Willem de Bruijn"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
qmi_wwan: Fix qmap header retrieval in qmimux_rx_fixup
qmi_wwan: Add support for Fibocom NL678 series
tls: Do not call sk_memcopy_from_iter with zero length
ipv6: tunnels: fix two use-after-free
Prevent overflow of sk_msg in sk_msg_clone()
packet: validate address length
net: netxen: fix a missing check and an uninitialized use
tcp: fix a race in inet_diag_dump_icsk()
MAINTAINERS: update cxgb4 and cxgb3 maintainer
ipv6: frags: Fix bogus skb->sk in reassembled packets
mscc: Configured MAC entries should be locked.
Mans Rullgard [Wed, 5 Dec 2018 13:52:47 +0000 (13:52 +0000)]
auxdisplay: charlcd: fix x/y command parsing
The x/y command parsing has been broken since commit 129957069e6a
("staging: panel: Fixed checkpatch warning about simple_strtoul()").
Commit b34050fadb86 ("auxdisplay: charlcd: Fix and clean up handling of
x/y commands") fixed some problems by rewriting the parsing code,
but also broke things further by removing the check for a complete
command before attempting to parse it. As a result, parsing is
terminated at the first x or y character.
This reinstates the check for a final semicolon. Whereas the original
code use strchr(), this is wasteful seeing as the semicolon is always
at the end of the buffer. Thus check this character directly instead.
Signed-off-by: Mans Rullgard <mans@mansr.com> Signed-off-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Yangtao Li [Wed, 12 Dec 2018 16:01:45 +0000 (11:01 -0500)]
serial/sunsu: fix refcount leak
The function of_find_node_by_path() acquires a reference to the node
returned by it and that reference needs to be dropped by its caller.
su_get_type() doesn't do that. The match node are used as an identifier
to compare against the current node, so we can directly drop the refcount
after getting the node from the path as it is not used as pointer.
Fix this by use a single variable and drop the refcount right after
of_find_node_by_path().
Signed-off-by: Yangtao Li <tiny.windzz@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Corentin Labbe [Tue, 11 Dec 2018 12:11:09 +0000 (12:11 +0000)]
sparc: Set "ARCH: sunxx" information on the same line
While checking boot log from SPARC qemu, I saw that the "ARCH: sunxx"
information was split on two different line.
This patchs merge both line together.
In the meantime, thoses information need to be printed via pr_info
since printk print them by default via the warning loglevel.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sparc: vdso: Drop implicit common-page-size linker flag
GNU linker's -z common-page-size's default value is based on the target
architecture. arch/sparc/vdso/Makefile sets it to the architecture
default, which is implicit and redundant. Drop it.
Vakul Garg [Fri, 21 Dec 2018 15:16:52 +0000 (15:16 +0000)]
tls: Do not call sk_memcopy_from_iter with zero length
In some conditions e.g. when tls_clone_plaintext_msg() returns -ENOSPC,
the number of bytes to be copied using subsequent function
sk_msg_memcopy_from_iter() becomes zero. This causes function
sk_msg_memcopy_from_iter() to fail which in turn causes tls_sw_sendmsg()
to return failure. To prevent it, do not call sk_msg_memcopy_from_iter()
when number of bytes to copy (indicated by 'try_to_copy') is zero.
Fixes: d829e9c4112b ("tls: convert to generic sk_msg interface") Signed-off-by: Vakul Garg <vakul.garg@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
This series includes some follow-up for the recently added skb extension.
The first patch addresses an unlikely race while adding skb extensions,
and the following two are just minor code clean-up.
v1 -> v2:
- be sure to flag the newly added extension as active in skb_ext_add()
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Paolo Abeni [Fri, 21 Dec 2018 18:03:15 +0000 (19:03 +0100)]
net: minor cleanup in skb_ext_add()
When the extension to be added is already present, the only
skb field we may need to update is 'extensions': we can reorder
the code and avoid a branch.
v1 -> v2:
- be sure to flag the newly added extension as active
Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Paolo Abeni [Fri, 21 Dec 2018 18:03:13 +0000 (19:03 +0100)]
net: fix possible user-after-free in skb_ext_add()
On cow we can free the old extension: we must avoid dereferencing
such extension after skb_ext_maybe_cow(). Since 'new' contents
are always equal to 'old' after the copy, we can fix the above
accessing the relevant data using 'new'.
Fixes: df5042f4c5b9 ("sk_buff: add skb extension infrastructure") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Acked-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Fri, 21 Dec 2018 17:22:24 +0000 (09:22 -0800)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Ingo Molnar:
"The biggest part is a series of reverts for the macro based GCC
inlining workarounds. It caused regressions in distro build and other
kernel tooling environments, and the GCC project was very receptive to
fixing the underlying inliner weaknesses - so as time ran out we
decided to do a reasonably straightforward revert of the patches. The
plan is to rely on the 'asm inline' GCC 9 feature, which might be
backported to GCC 8 and could thus become reasonably widely available
on modern distros.
Other than those reverts, there's misc fixes from all around the
place.
I wish our final x86 pull request for v4.20 was smaller..."
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
Revert "kbuild/Makefile: Prepare for using macros in inline assembly code to work around asm() related GCC inlining bugs"
Revert "x86/objtool: Use asm macros to work around GCC inlining bugs"
Revert "x86/refcount: Work around GCC inlining bug"
Revert "x86/alternatives: Macrofy lock prefixes to work around GCC inlining bugs"
Revert "x86/bug: Macrofy the BUG table section handling, to work around GCC inlining bugs"
Revert "x86/paravirt: Work around GCC inlining bugs when compiling paravirt ops"
Revert "x86/extable: Macrofy inline assembly code to work around GCC inlining bugs"
Revert "x86/cpufeature: Macrofy inline assembly code to work around GCC inlining bugs"
Revert "x86/jump-labels: Macrofy inline assembly code to work around GCC inlining bugs"
x86/mtrr: Don't copy uninitialized gentry fields back to userspace
x86/fsgsbase/64: Fix the base write helper functions
x86/mm/cpa: Fix cpa_flush_array() TLB invalidation
x86/vdso: Pass --eh-frame-hdr to the linker
x86/mm: Fix decoy address handling vs 32-bit builds
x86/intel_rdt: Ensure a CPU remains online for the region's pseudo-locking sequence
x86/dump_pagetables: Fix LDT remap address marker
x86/mm: Fix guard hole handling
Eric Dumazet [Fri, 21 Dec 2018 15:47:51 +0000 (07:47 -0800)]
ipv6: tunnels: fix two use-after-free
xfrm6_policy_check() might have re-allocated skb->head, we need
to reload ipv6 header pointer.
sysbot reported :
BUG: KASAN: use-after-free in __ipv6_addr_type+0x302/0x32f net/ipv6/addrconf_core.c:40
Read of size 4 at addr ffff888191b8cb70 by task syz-executor2/1304
Memory state around the buggy address: ffff888191b8ca00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff888191b8ca80: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
>ffff888191b8cb00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^ ffff888191b8cb80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ffff888191b8cc00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
Fixes: 0d3c703a9d17 ("ipv6: Cleanup IPv6 tunnel receive path") Fixes: ed1efb2aefbb ("ipv6: Add support for IPsec virtual tunnel interfaces") Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Fri, 21 Dec 2018 17:17:52 +0000 (09:17 -0800)]
Merge tag 'drm-fixes-2018-12-21' of git://anongit.freedesktop.org/drm/drm
Pull final drm fix from Daniel Vetter:
"Very calm week, so either everything perfect or everyone on holidays
already. Just one array_index_nospec patch, also for stable"
* tag 'drm-fixes-2018-12-21' of git://anongit.freedesktop.org/drm/drm:
drm/ioctl: Fix Spectre v1 vulnerabilities
Vakul Garg [Fri, 21 Dec 2018 15:55:46 +0000 (15:55 +0000)]
Prevent overflow of sk_msg in sk_msg_clone()
Fixed function sk_msg_clone() to prevent overflow of 'dst' while adding
pages in scatterlist entries. The overflow of 'dst' causes crash in kernel
tls module while doing record encryption.
Fixes: d829e9c4112b ("tls: convert to generic sk_msg interface") Signed-off-by: Vakul Garg <vakul.garg@nxp.com> Acked-by: John Fastabend <john.fastabend@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Willem de Bruijn [Fri, 21 Dec 2018 17:06:59 +0000 (12:06 -0500)]
packet: validate address length
Packet sockets with SOCK_DGRAM may pass an address for use in
dev_hard_header. Ensure that it is of sufficient length.
Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Fri, 21 Dec 2018 17:09:30 +0000 (09:09 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Pull input fixes from Dmitry Torokhov:
"Switching a few devices with Synaptics over to SMbus and disabling
SMbus on a couple devices with Elan touchpads as they need more
plumbing on PS/2 side"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: synaptics - enable SMBus for HP EliteBook 840 G4
Input: elantech - disable elan-i2c for P52 and P72
Input: synaptics - enable RMI on ThinkPad T560
Input: omap-keypad - fix idle configuration to not block SoC idle states
Linus Torvalds [Fri, 21 Dec 2018 17:05:28 +0000 (09:05 -0800)]
Merge tag 'gpio-v4.20-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio
Pull GPIO fixes from Linus Walleij:
"Hopefully last round of GPIO fixes.
The ACPI patch is pretty important for some laptop users, the rest is
driver-specific for embedded (mostly ARM) systems.
I took out one ACPI patch that wasn't critical enough because I
couldn't justify sending it at this point, and that is why the commit
date is today, but the patches have been in linux-next.
Sorry for not sending some of them earlier :(
Notice that we have a co-maintainer for GPIO now, Bartosz Golaszewski,
and he might jump in and make some pull requests at times when I am
off.
Summary:
- ACPI IRQ request deferral
- OMAP: revert deferred wakeup quirk
- MAX7301: fix DMA safe memory handling
- MVEBU: selective probe failure on missing clk"
* tag 'gpio-v4.20-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio:
gpio: mvebu: only fail on missing clk if pwm is actually to be used
gpio: max7301: fix driver for use with CONFIG_VMAP_STACK
gpio: gpio-omap: Revert deferred wakeup quirk handling for regressions
gpiolib-acpi: Only defer request_irq for GpioInt ACPI event handlers
Daniel Borkmann [Fri, 21 Dec 2018 13:04:46 +0000 (14:04 +0100)]
bpf: fix segfault in test_verifier selftest
Minor fallout from merge resolution, test_verifier was segfaulting
because the REJECT result was correct, but errstr was NULL. Properly
fix it as in 339bbff2d6e0.
Fixes: 339bbff2d6e0 ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next") Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Fri, 21 Dec 2018 16:56:31 +0000 (08:56 -0800)]
Merge tag '4.20-rc7-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
Pull smb3 fix from Steve French:
"An important smb3 fix for an regression to some servers introduced by
compounding optimization to rmdir.
This fix has been tested by multiple developers (including me) with
the usual private xfstesting, but also by the new cifs/smb3 "buildbot"
xfstest VMs (thank you Ronnie and Aurelien for good work on this
automation). The automated testing has been updated so that it will
catch problems like this in the future.
Note that Pavel discovered (very recently) some unrelated but
extremely important bugs in credit handling (smb3 flow control problem
that can lead to disconnects/reconnects) when compounding, that I
would have liked to send in ASAP but the complete testing of those two
fixes may not be done in time and have to wait for 4.21"
* tag '4.20-rc7-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
smb3: Fix rmdir compounding regression to strict servers
Eric Sandeen [Fri, 21 Dec 2018 16:42:50 +0000 (08:42 -0800)]
iomap: don't search past page end in iomap_is_partially_uptodate
iomap_is_partially_uptodate() is intended to check wither blocks within
the selected range of a not-uptodate page are uptodate; if the range we
care about is up to date, it's an optimization.
However, the iomap implementation continues to check all blocks up to
from+count, which is beyond the page, and can even be well beyond the
iop->uptodate bitmap.
I think the worst that will happen is that we may eventually find a zero
bit and return "not partially uptodate" when it would have otherwise
returned true, and skip the optimization. Still, it's clearly an invalid
memory access that must be fixed.
So: fix this by limiting the search to within the page as is done in the
non-iomap variant, block_is_partially_uptodate().
Zorro noticed thiswhen KASAN went off for 512 byte blocks on a 64k
page system:
BUG: KASAN: slab-out-of-bounds in iomap_is_partially_uptodate+0x1a0/0x1e0
Read of size 8 at addr ffff800120c3a318 by task fsstress/22337
Reported-by: Zorro Lang <zlang@redhat.com> Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Eric Sandeen <sandeen@sandeen.net> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Anup Patel [Tue, 4 Dec 2018 10:29:51 +0000 (15:59 +0530)]
RISC-V: Select GENERIC_SCHED_CLOCK for clocksource drivers
The riscv_timer driver can provide sched_clock using "rdtime"
instruction but to achieve this we require generic sched_clock
framework hence this patch selects GENERIC_SCHED_CLOCK for RISCV.
Palmer Dabbelt [Fri, 21 Dec 2018 16:15:39 +0000 (08:15 -0800)]
RISC-V: Move from EARLY_PRINTK to SBI earlycon
Now that we have earlycon support in the SBI console driver there is no
reason to have our arch-specific early printk support. This patch set
turns on SBI earlycon support and removes the old early printk.
Nick Kossifidis [Sun, 18 Nov 2018 00:06:56 +0000 (02:06 +0200)]
RISC-V: Update Kconfig to better handle CMDLINE
Added a menu to choose how the built-in command line will be
used and CMDLINE_EXTEND for compatibility with FDT code.
v2: Improved help messages, removed references to bootloader
and made them more descriptive. I also asked help from a
friend who's a language expert just in case.
v3: This time used the corrected text
v4: Copy the config strings from the arm32 port.
v5: Actually copy the config strings from the arm32 port.
David S. Miller [Fri, 21 Dec 2018 16:00:27 +0000 (08:00 -0800)]
Merge tag 'mlx5-XDP-100Mpps' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux
Saeed Mahameed says:
====================
mlx5-XDP-100Mpps
This series from Tariq, mainly adds the support of mlx5 Multi Packet WQE
(TX descriptor) - ConnectX-5 and above - for XDP TX, which allows us to
overcome the 70Mpps PCIe bottleneck of conventional TX queues (single TX
descriptor per packet), and achieve the 100Mpps milestone with the MPWQE
approach.
In the first five patches, Tariq did minor improvements to mlx5 tx path,
for better debug-ability and code structuring.
Next two patches lay down the foundation for MPWQE implementation to store
the in-flight XDP TX information for multiple packets of one descriptor
(WQE).
Next: Support Enhanced Multi-Packet TX WQE for XDP
In this patch we add support for the HW feature, which is supported
starting from ConnectX-5.
Performance:
Tested packet rate for UDP 64Byte multi-stream over ConnectX-5 NICs.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz
XDP_TX:
We see a huge gain on single port ConnectX-5, and reach the 100 Mpps
milestone.
* Single-port HCA:
Before: 70 Mpps
After: 100 Mpps (+42.8%)
* In both cases we tested traffic on one port and for now On Dual-port
HCAs we see only a small gain, we are working to overcome this
bottleneck, but for the moment only with experimental firmware on dual
port HCAs we can reach the wanted numbers as seen on Single-port HCAs.
XDP_REDIRECT:
Redirect from (A) ConnectX-5 to (B) ConnectX-5.
Due to a setup limitation, (A) and (B) are on different NUMA nodes,
so absolute performance numbers are not optimal.
- Note:
Below is the transmit rate of (B), not the redirect rate of (A)
which is in some cases higher.