1588 Single Step Timestamping code path uses a mutex to
enforce atomicity for two events:
- update of ptp single step register
- transmit ptp event packet
Before this patch the mutex was not initialized. This
caused unexpected crashes in the Tx function.
Fixes: c55211892f ("dpaa2-eth: support PTP Sync packet one-step timestamping")
Signed-off-by: Radu Bulie <radu-andrei.bulie@nxp.com>
Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Clang static analysis reports this representative problem
dpaa2-switch-flower.c:616:24: warning: The right operand of '=='
is a garbage value
tmp->cfg.vlan_id == vlan) {
^ ~~~~
vlan is set in dpaa2_switch_flower_parse_mirror_key(). However
this function can return success without setting vlan. So
change the default return to -EOPNOTSUPP.
Fixes: 0f3faece58 ("dpaa2-switch: add VLAN based mirroring")
Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When multiple containers are running in the environment and multiple
macvlan network port are configured in each container, a lot of martian
source prints will appear after martian_log is enabled. they are almost
the same, and printed by net_warn_ratelimited. Each arp message will
trigger this print on each network port.
Such as:
IPv4: martian source 173.254.95.16 from 173.254.100.109,
on dev eth0
ll header: 00000000: ff ff ff ff ff ff 40 00 ad fe 64 6d
08 06 ......@...dm..
IPv4: martian source 173.254.95.16 from 173.254.100.109,
on dev eth1
ll header: 00000000: ff ff ff ff ff ff 40 00 ad fe 64 6d
08 06 ......@...dm..
There is no description of this kind of source in the RFC1812.
Signed-off-by: Zhang Yunkai <zhang.yunkai@zte.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes the checkpatch.pl warning:
ERROR: code indent should use tabs where possible #3453: FILE: drivers/net/virtio_net.c:3453: ret = register_virtio_driver(&virtio_net_driver);$
Uneccessary newline was also removed making line 3453 now 3452.
Signed-off-by: Michael Catanzaro <mcatanzaro.kernel@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stefan Schmidt says:
====================
Only a single fix this time.
Miquel Raynal fixed the lifs/sifs periods in the ca82010 to take the actual
symbol duration time into account.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The function mt7531_phy_mode_supported in the DSA driver set supported
mode to PHY_INTERFACE_MODE_GMII instead of PHY_INTERFACE_MODE_INTERNAL
for the internal PHY, so this check breaks the PHY initialization:
mt7530 mdio-bus:00 wan (uninitialized): failed to connect to PHY: -EINVAL
Remove the check to make it work again.
Reported-by: Hauke Mehrtens <hauke@hauke-m.de>
Fixes: e40d2cca01 ("net: phy: add MediaTek Gigabit Ethernet PHY driver")
Signed-off-by: DENG Qingfang <dqfext@gmail.com>
Acked-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Tested-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The -ENODEV return value from xhci_check_args() is incorrectly changed
to -EINVAL in a couple places before propagated further.
xhci_check_args() returns 4 types of value, -ENODEV, -EINVAL, 1 and 0.
xhci_urb_enqueue and xhci_check_streams_endpoint return -EINVAL if
the return value of xhci_check_args <= 0.
This causes problems for example r8152_submit_rx, calling usb_submit_urb
in drivers/net/usb/r8152.c.
r8152_submit_rx will never get -ENODEV after submiting an urb when xHC
is halted because xhci_urb_enqueue returns -EINVAL in the very beginning.
[commit message and header edit -Mathias]
Fixes: 203a86613f ("xhci: Avoid NULL pointer deref when host dies.")
Cc: stable@vger.kernel.org
Signed-off-by: Hongyu Xie <xiehongyu1@kylinos.cn>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20220215123320.1253947-3-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When HCE(Host Controller Error) is set, it means an internal
error condition has been detected. Software needs to re-initialize
the HC, so add this check in xhci resume.
Cc: stable@vger.kernel.org
Signed-off-by: Puma Hsu <pumahsu@google.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20220215123320.1253947-2-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit e0082698b6 ("usb: dwc3: ulpi: conditionally resume ULPI PHY")
fixed an issue where ULPI transfers would timeout if any requests where
send to the phy sometime after init, giving it enough time to auto-suspend.
Commit e5f4ca3fce ("usb: dwc3: ulpi: Fix USB2.0 HS/FS/LS PHY suspend
regression") changed the behavior to instead of clearing the
DWC3_GUSB2PHYCFG_SUSPHY bit, add an extra sleep when it is set.
But on Bay Trail devices, when phy_set_mode() gets called during init,
this leads to errors like these:
[ 28.451522] tusb1210 dwc3.ulpi: error -110 writing val 0x01 to reg 0x0a
[ 28.464089] tusb1210 dwc3.ulpi: error -110 writing val 0x01 to reg 0x0a
Add "snps,dis_u2_susphy_quirk" to the settings for Bay Trail devices to
fix this. This restores the old behavior for Bay Trail devices, since
previously the DWC3_GUSB2PHYCFG_SUSPHY bit would get cleared on the first
ulpi_read/_write() and then was never set again.
Fixes: e5f4ca3fce ("usb: dwc3: ulpi: Fix USB2.0 HS/FS/LS PHY suspend regression")
Cc: stable@kernel.org
Cc: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20220213130524.18748-2-hdegoede@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The original workaround was added prior to commit e4788edc73 ("USB:
EHCI: Add alias for Broadcom INSNREG"). Now that brcm_insnreg exists in
struct ehci_regs we can use that instead of having a local definition.
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Link: https://lore.kernel.org/r/20220215000813.1779032-1-chris.packham@alliedtelesis.co.nz
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 'ret' local variables are often initialized to 0 but this value is
unused, thus we can kill those initializers...
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Link: https://lore.kernel.org/r/901b7478-45b6-d8b3-f5c6-555712485232@omp.ru
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The gh_error field if a glock holder is initialized to zero in
gfs2_holder_init(). When a locking operation fails, gh_error is set to
an error code; when it succeeds, the gh_error value is left unchanged.
The field isn't initialized in gfs2_holder_reinit(), which is a problem.
Instead of fixing that directly, initialize gh_error in gfs2_glock_nq().
That also obsoletes the assignment in do_flock().
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
This patch tries to fix the continual ABBA deadlocks we keep having
between the iopen and inode glocks. This switches the lock order in
gfs2_inode_lookup and gfs2_create_inode so the iopen glock is always
locked first.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
The gfs2 evict code tries to upgrade the iopen glock from SH to EX. If
the attempt to upgrade times out, gfs2 needs to tell dlm to cancel the
lock request or it can deadlock. We also need to wake up the process
waiting for the lock when dlm sends its AST back to gfs2.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Due to the asynchronous nature of the dlm api, when we request a pending
locking request to be canceled with dlm_unlock(DLM_LKF_CANCEL), the
locking request will either complete before it could be canceled, or the
cancellation will succeed. In either case, gdlm_ast will be called once
and the status will indicate the outcome of the locking request, with
-DLM_ECANCEL indicating a canceled request.
Inside dlm, when a locking request completes before its cancel request
could be processed, gdlm_ast will be called, but the lock will still be
considered busy until a DLM_MSG_CANCEL_REPLY message completes the
cancel request. During that time, successive dlm_lock() or dlm_unlock()
requests for that lock will return -EBUSY. In other words, waiting for
the gdlm_ast call before issuing the next locking request is not enough.
There is no way of waiting for a cancel request to actually complete,
either.
We rarely cancel locking requests, but when we do, we don't know when
the next locking request for that lock will occur. This means that any
dlm_lock() or dlm_unlock() call can potentially return -EBUSY. When
that happens, this patch simply repeats the request after a short pause.
This workaround could be improved upon by tracking for which dlm locks
cancel requests have been issued, but that isn't strictly necessary and
it would complicate the code. We haven't seen -EBUSY errors from dlm
without cancel requests.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
When gfs2_setattr_size() fails, it calls gfs2_rs_delete(ip, NULL) to get
rid of any reservations the inode may have. Instead, it should pass in
the inode's write count as the second parameter to allow
gfs2_rs_delete() to figure out if the inode has any writers left.
In a next step, there are two instances of gfs2_rs_delete(ip, NULL) left
where we know that there can be no other users of the inode. Replace
those with gfs2_rs_deltree(&ip->i_res) to avoid the unnecessary write
count check.
With that, gfs2_rs_delete() is only called with the inode's actual write
count, so get rid of the second parameter.
Fixes: a097dc7e24 ("GFS2: Make rgrp reservations part of the gfs2_inode structure")
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Before this patch, function read_rindex_entry called compute_bitstructs
before it allocated a glock for the rgrp. But if compute_bitstructs found
a problem with the rgrp, it called gfs2_consist_rgrpd, and that called
gfs2_dump_glock for rgd->rd_gl which had not yet been assigned.
read_rindex_entry
compute_bitstructs
gfs2_consist_rgrpd
gfs2_dump_glock <---------rgd->rd_gl was not set.
This patch changes read_rindex_entry so it assigns an rgrp glock before
calling compute_bitstructs so gfs2_dump_glock does not reference an
unassigned pointer. If an error is discovered, the glock must also be
put, so a new goto and label were added.
Reported-by: syzbot+c6fd14145e2f62ca0784@syzkaller.appspotmail.com
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
the other patches from the "[PATCH v6 0/9] Support Spi in
i2c-multi-instantiate driver" series.
-----BEGIN PGP SIGNATURE-----
iQFIBAABCAAyFiEEuvA7XScYQRpenhd+kuxHeUQDJ9wFAmH6vlwUHGhkZWdvZWRl
QHJlZGhhdC5jb20ACgkQkuxHeUQDJ9y3sAf+JoNPIvjcZLsrl1Xf5mk44c7Qg0BL
uknI7jj3FJem65uG4PKkB0SQ2Ysd5TfH0NQaUG3oiq+YqPJUjIm2qtsyBOT/cUWP
3EcCjX8/kYBtotTU3t16ciZCKXHW9zNvxTttUQJdDLEWyRLZ8RlRBTvcATdNK9DA
POA/f+Dh0PBhPFf4V5/XjylELNLAvzK+rYk2IdrGCSe6P3DOJU7M96qRHM7JQSym
tfxswPRwCDNr4A/3yRnI+N1XKwEdoEpiGIY2cXqoSzIePRjzvsxby9wLSuG4wp95
2/AUj4G9biQMsUSZoyeNU4henoNtUnUn8DOYa1ZaYxXl4nh04yICrqfLLg==
=njf6
-----END PGP SIGNATURE-----
Merge tag 'platform-drivers-x86-serial-multi-instantiate-1' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86 into for-next
This branch contains 5.17-rc1 + the SPI tree's spi-acpi-helpers tag +
the other patches from the "[PATCH v6 0/9] Support Spi in
i2c-multi-instantiate driver" series.
Merge series from Mika Westerberg <mika.westerberg@linux.intel.com>:
Based on discussion on the patch I sent some time ago here:
http://lists.infradead.org/pipermail/linux-mtd/2021-June/086867.html
it turns out that the preferred way to deal with the SPI flash controller
drivers is through SPI MEM which is part of Linux SPI subsystem.
This series does that for the intel-spi driver. This also renames the
driver to follow the convention used in the SPI subsystem. The first patch
improves the write protection handling to be slightly more safer. The
following two patches do the conversion itself. Note the Intel SPI flash
controller only allows commands such as read, write and so on and it
internally uses whatever addressing etc. it figured from the SFDP on the
flash device.
base-commit: e783362eb5
In the case like dmaengine which's not a dai but as a component, the
num_dai is zero, dmaengine component has the same component_of_node
as cpu dai, when cpu dai component is not ready, but dmaengine component
is ready, try to get cpu dai name, the snd_soc_get_dai_name() return
-EINVAL, not -EPROBE_DEFER, that cause below error:
asoc-simple-card <card name>: parse error -22
asoc-simple-card: probe of <card name> failed with error -22
The sound card failed to probe.
So this patch fixes the issue above by skipping the zero num_dai
component in searching dai name.
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Link: https://lore.kernel.org/r/1644491952-7457-1-git-send-email-shengjiu.wang@nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Don't populate the read-only const array rates on the stack but
instead it static. Also makes the object code a little smaller.
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20220214213223.65780-1-colin.i.king@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Get rid of the inflexible bigjoiner_linked_crtc pointer thing
and just track things as a bitmask of pipes instead. We can
also nuke the bigjoiner_slave boolean as the role of the pipe
can be determined from its position in the bitmask.
It might be possible to nuke the bigjoiner boolean as well
if we make encoder.compute_config() do the bitmask assignment
directly for the master pipe. But for now I left that alone so
that encoer.compute_config() will just flag the state as needing
bigjoiner, and the intel_atomic_check_bigjoiner() is still
responsible for determining the bitmask. But that may have to change
as the encoder may be in the best position to determine how
exactly we should populate the bitmask.
Most places that just looked at the single bigjoiner_linked_crtc
now iterate over the whole bitmask, eliminating the singular
slave pipe assumption.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-11-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>
Often using pipes is more convenient than crtc indices.
Convert the current for_each_intel_crtc_mask() to take a
pipe mask instead of a crtc index mask, and rename it to
for_each_intel_crtc_in_pipe_mask() to make it clear what
it does.
The current users of for_each_intel_crtc_mask() don't really
care which kind of mask we use, but for other uses a pipe
mask if better.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220203183823.22890-8-ville.syrjala@linux.intel.com
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>
* irq/print_chip:
: .
: Convert irqchip drivers that use the .name field as a topology
: description to the .irq_print_chip callback, which allows the
: name to be made dymanic. The irq_chip structures are then made
: 'const' in order to prevent further abuse.
: .
irqchip/versatile-fpga: Switch to dynamic chip name output
irqchip/ts4800: Switch to dynamic chip name output
irqchip/mvebu-pic: Switch to dynamic chip name output
irqchip/lpc32xx: Switch to dynamic chip name output
irqchip/gic: Switch to dynamic chip name output
genirq/debugfs: Use irq_print_chip() when provided by irqchip
genirq: Allow irq_chip registration functions to take a const irq_chip
irqdomain: Let irq_domain_set_{info,hwirq_and_chip} take a const irq_chip
Signed-off-by: Marc Zyngier <maz@kernel.org>
When clearing the group configurations, the driver fails to restore the
default setting for DSA 1.x based devices. Add defaults in
idxd_groups_clear_state() for traffic class configuration.
Fixes: ade8a86b51 ("dmaengine: idxd: Set defaults for GRPCFG traffic class")
Reported-by: Binuraj Ravindran <binuraj.ravindran@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/164304123369.824298.6952463420266592087.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Building tinyconfig with gcc (Debian 11.2.0-16) and assembler (Debian
2.37.90.20220207) the following build error shows up:
{standard input}: Assembler messages:
{standard input}:2088: Error: unrecognized opcode: `ptesync'
make[3]: *** [/builds/linux/scripts/Makefile.build:287: arch/powerpc/lib/sstep.o] Error 1
Add the 'ifdef CONFIG_PPC64' around the 'ptesync' in function
'emulate_update_regs()' to like it is in 'analyse_instr()'. Since it looks like
it got dropped inadvertently by commit 3cdfcbfd32 ("powerpc: Change
analyse_instr so it doesn't modify *regs").
A key detail is that analyse_instr() will never recognise lwsync or
ptesync on 32-bit (because of the existing ifdef), and as a result
emulate_update_regs() should never be called with an op specifying
either of those on 32-bit. So removing them from emulate_update_regs()
should be a nop in terms of runtime behaviour.
Fixes: 3cdfcbfd32 ("powerpc: Change analyse_instr so it doesn't modify *regs")
Cc: stable@vger.kernel.org # v4.14+
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
[mpe: Add last paragraph of change log mentioning analyse_instr() details]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220211005113.1361436-1-anders.roxell@linaro.org
Currently, `git status` lists the file as untracked by git, so tell git
to ignore it.
Fixes: aa3bc365ee ("powerpc/ps3: Add check for otheros image size")
Signed-off-by: Paul Menzel <pmenzel@molgen.mpg.de>
Acked-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220214065543.198992-1-pmenzel@molgen.mpg.de
Move the name output to the relevant callback, which allows us
some nice cleanups (mostly owing to the fact that the driver is
now DT only.
We also drop a random include directive from the ftintc010 driver.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-8-maz@kernel.org
Instead of overriding the name field, track the corresponding device
and use the relevant callback to output its name.
This allows us to make the irq_chip structure const.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Gregory CLEMENT <gregory.clement@bootlin.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-6-maz@kernel.org
Instead of overriding the name field with the device name, use
the relevant callback. This allows us to make the irq_chip structure
const.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-5-maz@kernel.org
The last dynamic aspect of the GIC's irq_chip structure is the
name that is associated to it.
Move the output of that name to the relevant callback, which
allows us to do a bit of cleanup and mark the structures const.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-4-maz@kernel.org
In order to let a const irqchip be fed to the irqchip layer, adjust
the various prototypes. An extra cast in irq_set_chip()() is required
to avoid a warning.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-3-maz@kernel.org
In order to let a const irqchip be fed to the irqchip layer, adjust
the various prototypes. An extra cast in irq_domain_set_hwirq_and_chip()
is required to avoid a warning.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220209162607.1118325-2-maz@kernel.org
mlx5 TX routines improvements
1) From Aya and Tariq, first 3 patches, Use the Max size of the TX descriptor
as advertised by the device and not the fixed value of 16 that the driver
always assumed, this is not a bug fix as all existing devices have Max value
larger than 16, but the series is necessary for future proofing the driver.
2) TX Synchronization improvements from Maxim, last 12 patches
Maxim Mikityanskiy Says:
=======================
mlx5e: Synchronize ndo_select_queue with configuration changes
The kernel can call ndo_select_queue at any time, and there is no direct
way to block it. The implementation of ndo_select_queue in mlx5e expects
the parameters to be consistent and may crash (invalid pointer, division
by zero) if they aren't.
There were attempts to partially fix some of the most frequent crashes,
see commit 846d6da1fc ("net/mlx5e: Fix division by 0 in
mlx5e_select_queue") and commit 84c8a87402 ("net/mlx5e: Fix division
by 0 in mlx5e_select_queue for representors"). However, they don't
address the issue completely.
This series introduces the proper synchronization mechanism between
mlx5e configuration and TX data path:
1. txq2sq updates are synchronized properly with ndo_start_xmit
(mlx5e_xmit). The TX queue is stopped when it configuration is being
updated, and memory barriers ensure the changes are visible before
restarting.
2. The set of parameters needed for mlx5e_select_queue is reduced, and
synchronization using RCU is implemented. This way, changes are
atomic, and the state in mlx5e_select_queue is always consistent.
3. A few optimizations are applied to the new implementation of
mlx5e_select_queue.
=======================
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEGhZs6bAKwk/OTgTpSD+KveBX+j4FAmILSJwACgkQSD+KveBX
+j6BjwgAj4tGNrIuR8wByLq7gFG+h8t/A80TkqxgNHFhYuF3u6bS6UnmymCAsM3F
G8jFott6Pu/4d7SQvksrtVQJvTTj+TinZFeLGNxYXLwm/Syk6Avbs2c02eDL1CbQ
fUE4HavgwpS+hp8xvdWOUH+i5b+e7I+iF6JC/K0LULHUhXaJ7BXiQi3qCm7DKwLm
gVdWzDHNBj73Xmxak71CH0b79qdIBMs0JlZjdizE++8vg/AZ8CY2G51UUTmjJk5h
YASXBigbmGAP5o30IWxjYaj51UQTmWmR5A1XhNIC4iVdWjFOPcz6k8M5EfB0ADq+
eXkQKGDiyMDddKAaor611K0n9H4Ogw==
=SDns
-----END PGP SIGNATURE-----
Merge tag 'mlx5-updates-2022-02-14' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux
Saeed Mahameed says:
====================
mlx5-updates-2022-02-14
mlx5 TX routines improvements
1) From Aya and Tariq, first 3 patches, Use the Max size of the TX descriptor
as advertised by the device and not the fixed value of 16 that the driver
always assumed, this is not a bug fix as all existing devices have Max value
larger than 16, but the series is necessary for future proofing the driver.
2) TX Synchronization improvements from Maxim, last 12 patches
Maxim Mikityanskiy Says:
=======================
mlx5e: Synchronize ndo_select_queue with configuration changes
The kernel can call ndo_select_queue at any time, and there is no direct
way to block it. The implementation of ndo_select_queue in mlx5e expects
the parameters to be consistent and may crash (invalid pointer, division
by zero) if they aren't.
There were attempts to partially fix some of the most frequent crashes,
see commit 846d6da1fc ("net/mlx5e: Fix division by 0 in
mlx5e_select_queue") and commit 84c8a87402 ("net/mlx5e: Fix division
by 0 in mlx5e_select_queue for representors"). However, they don't
address the issue completely.
This series introduces the proper synchronization mechanism between
mlx5e configuration and TX data path:
1. txq2sq updates are synchronized properly with ndo_start_xmit
(mlx5e_xmit). The TX queue is stopped when it configuration is being
updated, and memory barriers ensure the changes are visible before
restarting.
2. The set of parameters needed for mlx5e_select_queue is reduced, and
synchronization using RCU is implemented. This way, changes are
atomic, and the state in mlx5e_select_queue is always consistent.
3. A few optimizations are applied to the new implementation of
mlx5e_select_queue.
=======================
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Adjust the documentation to the new way how a PASID is being allocated,
freed and fixed up.
Based on a patch by Ashok Raj <ashok.raj@intel.com>
[ bp: Massage commit message, fix htmldocs build warning ]
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20220207230254.3342514-12-fenghua.yu@intel.com
The ENQCMD instruction implicitly accesses the PASID_MSR to fill in the
pasid field of the descriptor being submitted to an accelerator. But
there is no precise (and stable across kernel changes) point at which
the PASID_MSR is updated from the value for one task to the next.
Kernel code that uses accelerators must always use the ENQCMDS instruction
which does not access the PASID_MSR.
Check for use of the ENQCMD instruction in the kernel and warn on its
usage.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Link: https://lore.kernel.org/r/20220207230254.3342514-11-fenghua.yu@intel.com
The ENQCMD feature can only be used if CONFIG_INTEL_IOMMU_SVM is set.
Add X86_FEATURE_ENQCMD to the disabled features mask as appropriate so
that cpu_feature_enabled() can be used to check the feature.
[ bp: Massage commit message. ]
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20220207230254.3342514-10-fenghua.yu@intel.com
All tasks start with PASID state disabled. This means that the first
time they execute an ENQCMD instruction they will take a #GP fault.
Modify the #GP fault handler to check if the "mm" for the task has
already been allocated a PASID. If so, try to fix the #GP fault by
loading the IA32_PASID MSR.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20220207230254.3342514-9-fenghua.yu@intel.com
Add a new single bit field to the task structure to track whether this task
has initialized the IA32_PASID MSR to the mm's PASID.
Initialize the field to zero when creating a new task with fork/clone.
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Co-developed-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20220207230254.3342514-8-fenghua.yu@intel.com
