| Age | Commit message (Collapse) | Author |
|---|
|
Instead of tracking wmem_queued and sk_mem_charge by incrementing
in the verdict SK_REDIRECT paths and decrementing in the tx work
path use skb_set_owner_w and sock_writeable helpers. This solves
a few issues with the current code. First, in SK_REDIRECT inc on
sk_wmem_queued and sk_mem_charge were being done without the peers
sock lock being held. Under stress this can result in accounting
errors when tx work and/or multiple verdict decisions are working
on the peer psock.
Additionally, this cleans up the code because we can rely on the
default destructor to decrement memory accounting on kfree_skb. Also
this will trigger sk_write_space when space becomes available on
kfree_skb() which wasn't happening before and prevent __sk_free
from being called until all in-flight packets are completed.
Fixes: 174a79ff9515 ("bpf: sockmap with sk redirect support")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Eric Dumazet says:
====================
net: ubuf_info.refcnt conversion
Yet another atomic_t -> refcount_t conversion, split in two patches.
First patch prepares the automatic conversion done in the second patch.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.
v2: added the change in drivers/vhost/net.c as spotted
by Willem.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
In order to convert this atomic_t refcnt to refcount_t,
we need to init the refcount to one to not trigger
a 0 -> 1 transition.
This also removes one atomic operation in fast path.
v2: removed dead code in sock_zerocopy_put_abort()
as suggested by Willem.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Similarly to how we configure the RSB (Receive Status Block) we also
need to set the TSB (Transmit Status Block) based on the host endian.
This was missing from the commit indicated below.
Fixes: 389a06bc534e ("net: systemport: Set correct RSB endian bits based on host")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Ivan Delalande says:
====================
inet_diag: report TCP MD5 signing keys and addresses
Allow userspace to retrieve MD5 signature keys and addresses configured
on TCP sockets through inet_diag.
Thanks to Eric Dumazet and Stephen Hemminger for their useful
explanations and feedback.
v5: - memset the whole netlink payload after it has been nla_reserve-d
in tcp_diag_put_md5sig (a third memset had to be added for
tcpm_key so we might as well have just one for entire region).
- move the nla_total_size call from inet_sk_attr_size to the
idiag_get_aux_size defined by protocols as they could add multiple
netlink attributes,
- add check for net_admin in tcp_diag_get_aux_size.
v4: - add new struct tcp_diag_md5sig to report the data instead of
tcp_md5sig to avoid wasting 112 bytes on every tcpm_addr,
- memset tcpm_addr on IPv4 addresses to avoid leaks,
- style fix in inet_diag_dump_one_icsk.
v3: - rename inet_diag_*md5sig in tcp_diag.c to tcp_diag_* for
consistency,
- don't lock the socket in tcp_diag_put_md5sig,
- add checks on md5sig_count in tcp_diag_put_md5sig to not create
the netlink attribute if the list is empty, and to avoid overflows
or memory leaks if the list has changed in the meantime.
v2: - move changes to tcp_diag.c and extend inet_diag_handler to allow
protocols to provide additional data on INET_DIAG_INFO,
- lock socket before calling tcp_diag_put_md5sig.
I also have a patch for iproute2/ss to test this change, making it print
this new attribute. I'm planning to polish and send it if this series
gets applied.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Report TCP MD5 (RFC2385) signing keys, addresses and address prefixes to
processes with CAP_NET_ADMIN requesting INET_DIAG_INFO. Currently it is
not possible to retrieve these from the kernel once they have been
configured on sockets.
Signed-off-by: Ivan Delalande <colona@arista.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Extend inet_diag_handler to allow individual protocols to report
additional data on INET_DIAG_INFO through idiag_get_aux. The size
can be dynamic and is computed by idiag_get_aux_size.
Signed-off-by: Ivan Delalande <colona@arista.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Grepping for "sizeof\(.+\) / sizeof\(" found this as one of the first
candidates.
Maybe a coccinelle can catch all of those.
Signed-off-by: Thomas Meyer <thomas@m3y3r.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
priv is being assigned but is never used, so remove it.
Cleans up clang build warning:
"warning: Value stored to 'priv' is never read"
Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial implementation")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
code size
Don't populate the array bcm7xxx_suspend_cfg A on the stack, instead
make it static. Makes the object code smaller by over 300 bytes:
Before:
text data bss dec hex filename
6351 8146 0 14497 38a1 drivers/net/phy/bcm7xxx.o
After:
text data bss dec hex filename
5986 8210 0 14196 3774 drivers/net/phy/bcm7xxx.o
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Don't populate the arrays port_ids on the stack, instead make them static.
Makes the object code smaller by over 700 bytes:
Before:
text data bss dec hex filename
28785 5832 192 34809 87f9 fman.o
After:
text data bss dec hex filename
27921 5992 192 34105 8539 fman.o
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Three cases of simple overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Excess of seafood or something happened while I cooked the commit
adding RB tree to inetpeer.
Of course, RCU rules need to be respected or bad things can happen.
In this particular loop, we need to read *pp once per iteration, not
twice.
Fixes: b145425f269a ("inetpeer: remove AVL implementation in favor of RB tree")
Reported-by: John Sperbeck <jsperbeck@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Olof Johansson:
"A couple of late-arriving fixes before final 4.13:
- A few reverts of DT bindings on Allwinner for their ethernet
driver. Discussion didn't converge, and since bindings are
considered ABI it makes sense to revert instead of having to
support two bindings long-term.
- A fix to enumerate GPIOs properly on Marvell Armada AP806"
* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
arm64: dts: marvell: fix number of GPIOs in Armada AP806 description
arm: dts: sunxi: Revert EMAC changes
arm64: dts: allwinner: Revert EMAC changes
dt-bindings: net: Revert sun8i dwmac binding
|
|
Sometimes root file system is deployed on SD card. It's
proper to build eSDHC driver into kernel.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux into next/dt
DT for 4.14
- sama5d2: add classD, ISC and QSPI
- New SoM and board: sama5d27 SoM1
- at91sam9g45: add AC97
* tag 'at91-ab-4.14-dt2' of git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux:
ARM: dts: at91: at91sam9g45: add AC97
ARM: dts: at91: sama5d2_xplained: add pin muxing and enable classd
ARM: dts: at91: sama5d2: add classd nodes
ARM: dts: at91: sama5d2_xplained: use pin macros instead of numbers
ARM: dts: at91: at91-sama5d27_som1_ek: Add sama5d27 SoM1 EK support
ARM: dts: at91: at91-sama5d27_som1: add sama5d27 SoM1 support
ARM: dts: at91: sama5d2: add isc node
ARM: dts: at91: sama5d2: add QSPI nodes
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
next/dt64
mvebu dt64 for 4.14 (part 4)
Adding more resources on the network controller ppv2.2 on Armada 7K/8K
allowing to use last improvement introduced in the driver.
Also enabling more network ports on the mcbin (A8K base board)
* tag 'mvebu-dt64-4.14-4' of git://git.infradead.org/linux-mvebu:
arm64: dts: marvell: mcbin: enable more networking ports
arm64: dts: marvell: add a reference to the sysctrl syscon in the ppv2 node
arm64: dts: marvell: add TX interrupts for PPv2.2
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
https://git.kernel.org/pub/scm/linux/kernel/git/sunxi/linux into next/dt64
Allwinner fixes for 4.13, take 3
This is a revert of the EMAC bindings. The discussion has not settled down
yet on a proper representation of the PHY, and therefore we cannot commit
to a binding yet
* tag 'sunxi-fixes-for-4.13-3' of https://git.kernel.org/pub/scm/linux/kernel/git/sunxi/linux:
arm: dts: sunxi: Revert EMAC changes
arm64: dts: allwinner: Revert EMAC changes
dt-bindings: net: Revert sun8i dwmac binding
arm64: allwinner: h5: fix pinctrl IRQs
arm64: allwinner: a64: sopine: add missing ethernet0 alias
arm64: allwinner: a64: pine64: add missing ethernet0 alias
arm64: allwinner: a64: bananapi-m64: add missing ethernet0 alias
Signed-off-by: Olof Johansson <olof@lixom.net>
Conflicts:
arch/arm64/boot/dts/allwinner/sun50i-a64-bananapi-m64.dts
arch/arm64/boot/dts/allwinner/sun50i-a64-pine64.dts
arch/arm64/boot/dts/allwinner/sun50i-a64-sopine-baseboard.dts
|
|
https://git.kernel.org/pub/scm/linux/kernel/git/sunxi/linux into next/dt
Allwinner fixes for 4.13, take 3
This is a revert of the EMAC bindings. The discussion has not settled down
yet on a proper representation of the PHY, and therefore we cannot commit
to a binding yet
* tag 'sunxi-fixes-for-4.13-3' of https://git.kernel.org/pub/scm/linux/kernel/git/sunxi/linux:
arm: dts: sunxi: Revert EMAC changes
arm64: dts: allwinner: Revert EMAC changes
dt-bindings: net: Revert sun8i dwmac binding
arm64: allwinner: h5: fix pinctrl IRQs
arm64: allwinner: a64: sopine: add missing ethernet0 alias
arm64: allwinner: a64: pine64: add missing ethernet0 alias
arm64: allwinner: a64: bananapi-m64: add missing ethernet0 alias
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
mvebu fixes for 4.13 (part 3)
Fix number of GPIOs in AP806 description for Armada 7K/8K
* tag 'mvebu-fixes-4.13-3' of git://git.infradead.org/linux-mvebu:
arm64: dts: marvell: fix number of GPIOs in Armada AP806 description
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
Changing the audio sample rate on the SolidRun Cubox disrupts the video
output. The Si5351 provides both the video clock (using PLLA on output
0) and the audio clock (using PLLB on output 2).
When the rate of clock output 2 is changed, it reconfigures PLLB, which
results in both PLLA and PLLB being reset. The reset of PLLA causes
clock output 0 to be disrupted, thereby causing a loss of sync by the
attached display device.
Hence, each time the audio sample rate changes (eg, when a video player
starts up, or when starting to play music) the video display momentarily
blanks while the Si5351 settles down. Prior to the commit below, this
behaviour did not happen.
Fix this by only resetting only the PLL which has been changed.
Fixes: 6dc669a22c77 ("clk: si5351: Add PLL soft reset")
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Acked-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
In certain circumstances rpmsg devices needs to acquire a handle to the
ancestor remoteproc instance, e.g. to invoke rproc_report_crash() when a
fatal error is detected. Introduce an interface that walks the device
tree in search for a remoteproc instance and return this.
Tested-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
|
|
Since gclk (generated-clk) is now able to determine the rate of the
audio_pll, there is no need for classd to have a direct phandle to the
audio_pll while already having a phandle to gclk.
Thus, remove all mentions to aclk in classd driver and update macros and
variable names.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Mark Brown <broonie@kernel.org>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
Since gclk (generated-clk) is now able to determine the rate of the
audio_pll, there is no need for classd to have a direct phandle to the
audio_pll while already having a phandle to gclk.
This binding is used by no board in mainline so it is safe to be
modified.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Mark Brown <broonie@kernel.org>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
This allows gclk to determine audio_pll rate and set the parent rate
accordingly.
However, there are multiple children clocks that could technically
change the rate of audio_pll (via gck). With the rate locking, the first
consumer to enable the clock will be the one definitely setting the rate
of the clock.
Since audio IPs are most likely to request the same rate, we enforce
that the only clks able to modify gck rate are those of audio IPs.
To remain consistent, we deny other clocks to be children of audio_pll.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
The way to find the best_diff and do the appropriate process afterwards
can be re-used.
This patch prepares the driver for an upcoming patch that will allow
clk_generated to determine the rate of the audio_pll.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
This new clock driver set allows to have a fractional divided clock that
would generate a precise clock particularly suitable for audio
applications.
The main audio pll clock has two children clocks: one that is connected
to the PMC, the other that can directly drive a pad. As these two routes
have different enable bits and different dividers and divider formulas,
they are handled by two different drivers. Each of them could modify the
rate of the main audio pll parent.
The main audio pll clock can output 620MHz to 700MHz.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
This new clock driver set allows to have a fractional divided clock that
would generate a precise clock particularly suitable for audio
applications.
The main audio pll clock has two children clocks: one that is connected
to the PMC, the other that can directly drive a pad. As these two routes
have different enable bits and different dividers and divider formulas,
they are handled by two different drivers.
This adds the audio plls (frac, pad and pmc) to the compatible list of
at91 clocks in DT binding.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Rob Herring <robh@kernel.org>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
The driver requests the current clk rate of each of its parent clocks to
decide whether a clock rate is suitable or not. It does not request
determine_rate from a parent clock which could request a rate change in
parent clock (i.e. there is no parent rate propagation).
We know the rate we want (passed along req argument of the function) and
the parent clock rate, thus we know the closest rounded divisor, we
don't need to iterate over the available divisors to find the best one
for a given clock.
Signed-off-by: Quentin Schulz <quentin.schulz@free-electrons.com>
Acked-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
The ssr_notifiers variable should be local, so add the missing static
storage classifier.
Fixes: 1e140df04965 ("remoteproc: qcom: Add support for SSR notifications")
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
|
|
In addition to using GLINK for communication with the RPM it can be
used ontop of SMEM for communicating with remoteprocs, extend the
binding to also describe this case and reference the GLINK binding from
the affected remoteproc bindings.
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
|
|
Introduce the GLINK subdev, which allows the definition of a GLINK edge
as child of a remoteproc.
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
|
|
Older compilers think val may be used uninitialized:
arch/powerpc/lib/sstep.c: In function 'emulate_loadstore':
arch/powerpc/lib/sstep.c:2758:23: error: 'val' may be used uninitialized in this function
We know better, but initialise val to 0 to avoid breaking the build.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux
Pull i2c fixes from Wolfram Sang:
"The ismt driver had a problem with a rarely used transaction type and
the designware driver was made even more robust against non standard
ACPI tables"
* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
i2c: designware: Round down ACPI provided clk to nearest supported clk
i2c: ismt: Return EMSGSIZE for block reads with bogus length
i2c: ismt: Don't duplicate the receive length for block reads
|
|
Fix various typos and whitespace errors:
s/Synopsis/Synopsys/
s/Designware/DesignWare/
s/Keystine/Keystone/
s/gpio/GPIO/
s/pcie/PCIe/
s/phy/PHY/
s/confgiruation/configuration/
No functional change intended.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
The "res" variable in pci_resource_io() is never used. Remove it.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
The kernel-doc comments don't match the arguments, so fix the comments.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
|
|
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
|
|
It fails to build once we introduce the ARCH_MB86S7X Kconfig symbol:
drivers/clk/clk-mb86s7x.c:27:10: fatal error: soc/mb86s7x/scb_mhu.h: No such file or directory
#include <soc/mb86s7x/scb_mhu.h>
^~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
And when commenting out that line, we get:
drivers/clk/clk-mb86s7x.c: In function 'crg_gate_control':
drivers/clk/clk-mb86s7x.c:72:8: error: implicit declaration of function 'mb86s7x_send_packet' [-Werror=implicit-function-declaration]
ret = mb86s7x_send_packet(CMD_PERI_CLOCK_GATE_SET_REQ,
^~~~~~~~~~~~~~~~~~~
drivers/clk/clk-mb86s7x.c:72:28: error: 'CMD_PERI_CLOCK_GATE_SET_REQ' undeclared (first use in this function)
ret = mb86s7x_send_packet(CMD_PERI_CLOCK_GATE_SET_REQ,
^~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/clk/clk-mb86s7x.c:72:28: note: each undeclared identifier is reported only once for each function it appears in
drivers/clk/clk-mb86s7x.c: In function 'crg_rate_control':
drivers/clk/clk-mb86s7x.c:116:10: error: 'CMD_PERI_CLOCK_RATE_SET_REQ' undeclared (first use in this function)
code = CMD_PERI_CLOCK_RATE_SET_REQ;
^~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/clk/clk-mb86s7x.c:121:10: error: 'CMD_PERI_CLOCK_RATE_GET_REQ' undeclared (first use in this function); did you mean 'CMD_PERI_CLOCK_RATE_SET_REQ'?
code = CMD_PERI_CLOCK_RATE_GET_REQ;
^~~~~~~~~~~~~~~~~~~~~~~~~~~
CMD_PERI_CLOCK_RATE_SET_REQ
drivers/clk/clk-mb86s7x.c: In function 'mhu_cluster_rate':
drivers/clk/clk-mb86s7x.c:276:10: error: 'CMD_CPU_CLOCK_RATE_GET_REQ' undeclared (first use in this function)
code = CMD_CPU_CLOCK_RATE_GET_REQ;
^~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/clk/clk-mb86s7x.c:278:10: error: 'CMD_CPU_CLOCK_RATE_SET_REQ' undeclared (first use in this function); did you mean 'CMD_CPU_CLOCK_RATE_GET_REQ'?
code = CMD_CPU_CLOCK_RATE_SET_REQ;
^~~~~~~~~~~~~~~~~~~~~~~~~~
CMD_CPU_CLOCK_RATE_GET_REQ
cc1: some warnings being treated as errors
scripts/Makefile.build:302: recipe for target
'drivers/clk/clk-mb86s7x.o' failed
make[2]: *** [drivers/clk/clk-mb86s7x.o] Error 1
Remove the driver for now.
Signed-off-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
|
|
The FMR_OF_LAST flag is set on the last fsmap record being returned for
the dataset requested, contrary to what the header file says. Fix the
docs to reflect the behavior of all fsmap implementations.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
|
|
In Christoph's patch to refactor xfs_bmse_merge, the updated rmap code
does more work than it needs to (because map-extent auto-merges
records). Remove the unnecessary unmap and save ourselves a deferred
op.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
|
|
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
When calling into _xfs_log_force{,_lsn}() with a pointer
to log_flushed variable, log_flushed will be set to 1 if:
1. xlog_sync() is called to flush the active log buffer
AND/OR
2. xlog_wait() is called to wait on a syncing log buffers
xfs_file_fsync() checks the value of log_flushed after
_xfs_log_force_lsn() call to optimize away an explicit
PREFLUSH request to the data block device after writing
out all the file's pages to disk.
This optimization is incorrect in the following sequence of events:
Task A Task B
-------------------------------------------------------
xfs_file_fsync()
_xfs_log_force_lsn()
xlog_sync()
[submit PREFLUSH]
xfs_file_fsync()
file_write_and_wait_range()
[submit WRITE X]
[endio WRITE X]
_xfs_log_force_lsn()
xlog_wait()
[endio PREFLUSH]
The write X is not guarantied to be on persistent storage
when PREFLUSH request in completed, because write A was submitted
after the PREFLUSH request, but xfs_file_fsync() of task A will
be notified of log_flushed=1 and will skip explicit flush.
If the system crashes after fsync of task A, write X may not be
present on disk after reboot.
This bug was discovered and demonstrated using Josef Bacik's
dm-log-writes target, which can be used to record block io operations
and then replay a subset of these operations onto the target device.
The test goes something like this:
- Use fsx to execute ops of a file and record ops on log device
- Every now and then fsync the file, store md5 of file and mark
the location in the log
- Then replay log onto device for each mark, mount fs and compare
md5 of file to stored value
Cc: Christoph Hellwig <hch@lst.de>
Cc: Josef Bacik <jbacik@fb.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
Currently flag switching can be used to easily crash the kernel. Disable
the per-inode DAX flag until that is sorted out.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
Use the existing functionality instead of directly poking into the extent
list.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
This avoids poking into the internals of the extent list. Also return
the number of extents as the return value instead of an additional
by reference argument, and make it available to callers outside of
xfs_bmap_util.c
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
This abstracts the function away from details of the low-level extent
list implementation.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
This abstracts the function away from details of the low-level extent
list implementation.
Note that it seems like the previous implementation of rmap for
the merge case was completely broken, but it no seems appear to
trigger that.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
|
For the first right move we need to look up next_fsb. That means
our last fsb that contains next_fsb must also be the current extent,
so take advantage of that by moving the code around a bit.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
|
