Jon Hunter [Tue, 5 May 2015 14:17:59 +0000 (15:17 +0100)]
serial: tegra: Correct error handling on DMA setup
Function tegra_uart_dma_channel_allocate() does not check that
dma_map_single() mapped the DMA buffer correctly. Add a check for this
and appropriate error handling.
Furthermore, if dmaengine_slave_config() (called by
tegra_uart_dma_channel_allocate()) fails, then memory allocated/mapped
is not freed/unmapped. Therefore, call tegra_uart_dma_channel_free()
instead of just dma_release_channel() if dmaengine_slave_config() fails.
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jon Hunter [Tue, 5 May 2015 14:17:58 +0000 (15:17 +0100)]
serial: tegra: Correct shutdown of UARTs
There are two issues in the shutdown path of the UARTs which are:
1. The function tegra_uart_shutdown() calls tegra_uart_flush_buffer()
to stop DMA TX transfers. However, tegra_uart_flush_buffer() is
called after the DMA channels have already been freed and so actually
does nothing.
2. The function that frees the DMA channels
(tegra_uart_dma_channel_free()), unmaps the dma buffer before
freeing the DMA channel and does not ensure the DMA has been
stopped.
Resolve this by fixing the code in tegra_uart_dma_channel_free() to
ensure the DMA is stopped, free the DMA channel and then unmap the DMA
buffer. Finally, remove the unnecessary call to tegra_uart_flush_buffer().
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jon Hunter [Tue, 5 May 2015 14:17:56 +0000 (15:17 +0100)]
serial: tegra: Use unsigned types for RX and TX byte counts
The function tty_insert_flip_string() takes an argument "size" which is
of type size_t. This is an unsigned type. Update the count,
rx_bytes_requested and tx_bytes_requested in the tegra serial driver to
be unsigned integers so that an unsigned type is passed to
tty_insert_flip_string().
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
serial: tegra: handle race condition on uart rx side
The tegra serial driver has two paths through which receive data is
copied up to the tty layer. These are:
1. DMA completion callback
2. UART RX interrupt
A UART RX interrupt occurs for either RX_TIMEOUT (data has been sitting
in the Rx FIFO for more than 4 character times without being read
because there is not enough data to reach the trigger level), End of
Receive Data event (receiver detects that data stops coming in for more
than 4 character times) or a receive error.
In the RX interrupt path, the following happens ...
- All RX DMA transfers are stopped
- Any data in the DMA buffer and RX FIFO are copied up to the tty layer.
- DMA is restarted/primed for the RX path
In the DMA completion callback, the DMA buffer is copied up to the tty
layer but there is no check to see if the RX interrupt could have
occurred between the DMA interrupt firing the the DMA callback running.
Hence, if a RX interrupt was to occur shortly after the DMA completion
interrupt, it is possible that the RX interrupt path has already copied
the DMA buffer before the DMA callback has been called. Therefore, when
the DMA callback is called, if the DMA is already in-progress, then this
indicates that the UART RX interrupt has already occurred and there is
nothing to do in the DMA callback. This race condition can cause
duplicated data to be received.
Signed-off-by: Shardar Shariff Md <smohammed@nvidia.com>
[jonathanh@nvidia.com: Moved async_tx_ack() call to after check to see
if DMA has completed because if the DMA is in progress we do not need
to ACK yet. Changed the print from dev_info to dev_debug. Updated
changelog to add more commentary on the race condition based upon
feedback from author.] Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
serial: tegra: check the count and read if any from dma
It is only necessary to read data from the dma buffer when the count
value is non-zero and hence, tegra_uart_copy_rx_to_tty() so only be
called when this is the case.
Although, this was being tested for in two places, there is a third
place where this was not tested. However, instead of adding another
if-statement prior to calling tegra_uart_copy_rx_to_tty(), move the test
inside the function.
Signed-off-by: Shardar Shariff Md <smohammed@nvidia.com>
[jonathanh@nvidia.com: Re-worked patch to move the check for the count
value inside the function tegra_uart_copy_rx_to_tty(). Updated
changelog with more commentary.] Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jon Hunter [Tue, 5 May 2015 14:17:53 +0000 (15:17 +0100)]
serial: tegra: Add delay after enabling FIFO mode
For all tegra devices (up to t210), there is a hardware issue that
requires software to wait for 3 UART clock periods after enabling
the TX fifo, otherwise data could be lost.
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jon Hunter [Tue, 5 May 2015 14:17:52 +0000 (15:17 +0100)]
serial: tegra: Correct delay after TX flush
For all tegra devices (up to t210), there is a hardware issue that
requires software to wait for 32 UART clock periods for the flush
to propagate otherwise TX data could be post. Add a helper function
to wait for N UART clock periods and update delay following FIFO
flush to be 32 UART clock cycles.
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jan Kiszka [Tue, 5 May 2015 06:26:27 +0000 (08:26 +0200)]
serial: 8250: Do nothing if nr_uarts=0
When nr_uarts was set to 0 (via config or 8250_core.nr_uarts), we crash
early on x86 because serial8250_isa_init_ports dereferences base_ops
which remains NULL. In fact, there is nothing to do for all the callers
of serial8250_isa_init_ports if there are no uarts.
Based on suggestions by Peter Hurley.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
serial: sh-sci: Don't set SCLSR bits in the SCxSR error mask
error_mask is the union of all error indicating bits in the SCxSR
register, while overrun_mask may apply to a different register (SCLSR),
depending on the SCI variant.
Hence overrun_mask should only be ORed into error_mask if it applies to
the SCxSR register.
serial: sh-sci: Use the correct register for overrun checks
The various SCI implementations use 3 different methods to signal
overrun errors:
- Bit SCI_ORER in register SCxSR on SCI,
- Bit SCIFA_ORER in register SCxSR on SCIFA and SCIFB, and SCIF on
SH7705/SH7720/SH7721,
- Bit SCLSR_ORER in (optional!) register SCLSR on (H)SCIF.
However:
1. sci_handle_fifo_overrun()
a. handles (H)SCIF and SCIFA/SCIFB only,
b. treats SCIF on SH7705/SH7720/SH7721 incorrectly,
2. sci_mpxed_interrupt()
a. treats SCIF on SH7705/SH7720/SH7721 incorrectly,
b. ignores that not all SCIFs have the SCLSR register, causing
"Invalid register access" WARN()ings.
To fix the above:
1. Determine and store the correct register enum during
initialization,
2. Replace the duplicated buggy switch statements by using the stored
register enum,
3. Add the missing existence check to sci_mpxed_interrupt().
serial: sh-sci: Replace hardcoded overrun bit values
Add the missing overrun bit definition for (H)SCIF.
Replace overrun_bit by overrun_mask, so we can use the existing
defines instead of hardcoded values.
serial: sh-sci: Replace hardcoded values in SCxSR_*_CLEAR macros
Add the missing overrun error bit in SCxSR on SCIFA/SCIFB and SCIF on
SH7705/SH7720/SH7721.
Document what the corresponding bit(s) on plain SCIF are used for.
Sort the components of SCIF_DEFAULT_ERROR_MASK by reverse definition
order.
Replace the hardcoded values in the SCxSR_*_CLEAR macros by proper
defines. Use bit masks (negations of sets of bits) to make it more
obvious which bits are being cleared.
Assembler output (on sh) was compared before and after this commit:
- For the first branch of the big "#if defined(...) || ..." construct,
the code has changed slightly, as 32-bit bitmasks can be loaded in a
single instruction, unlike the old large 16-bit constants (the SCxSR
register is 16 bit, so we don't care about the top 16 bits),
- For the second branch, the generated code is identical.
serial: sh-sci: Move private definitions to private header file
Move private register definitions and enums from the public
<linux/serial_sci.h> header file to the driver private "sh-sci.h" header
file.
The common Serial Control Register definitions are left in the public
header file, as they're needed to fill in plat_sci_port.scscr on legacy
systems not using DT.
Qipan Li [Wed, 29 Apr 2015 06:45:09 +0000 (06:45 +0000)]
serial: sirf: use uart_port's fifosize for fifo related operation
In SiRF platform, there are different fifo size of uart and usp,
with the fifosize configuration changes in different chips, we
can not use port line to decide how to check FIFO full,empty and
level.
There is a direct mapping between FIFO HW register layout with
fifo size, so move to use fifosize as the input to check fifo
status.
Signed-off-by: Qipan Li <Qipan.Li@csr.com> Signed-off-by: Barry Song <Baohua.Song@csr.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
at the same, enlarge the max port number according to the chip with the most
UART.
Signed-off-by: Qipan Li <Qipan.Li@csr.com> Signed-off-by: Barry Song <Baohua.Song@csr.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Eddie Huang [Tue, 28 Apr 2015 13:40:33 +0000 (21:40 +0800)]
tty: serial: 8250_mtk: Add earlycon
Add 8250 MTK UART driver to support earlycon device tree.
Earlycon take effect by
add "earlycon" in kernel boot argument
add "linux,sdtout-path" property in device tree file
The mtk 8250 needs two clocks, one for providing the baudrate and
one that needs to be enabled for register accesses. The latter has
not been supported, this patch adds support for it. It is optional
for now since not all SoCs provide a bus clock.
tty: serial: 8250_mtk: use pm_runtime callbacks for enabling
The pm_runtime callbacks already enable and disable the device.
Use them in probe() and remove() instead of duplicating the
code. This allows us to concentrate more code for enabling/disabling
the UART in a single place.
When a struct device * is present clk_get should be used rather
than of_clk_get. Use the devm variant of this function to be able to
drop the clk_put in the error and remove pathes. While at it fix
a wrong error message.
Firo Yang [Sun, 26 Apr 2015 10:46:06 +0000 (18:46 +0800)]
ARM: meson: serial: convert iounmap to devm_iounmap
The function meson_uart_release_port() inappropriately try to
iounmap() a resource managed by devm_ioremap_nocache().
The function meson_uart_release_port() maybe called by uart_ioctl()
that means meson_uart_release_port() is not called from within a
probe or remove function, for safety, I convert iounmap() to
devm_iounmap().
Signed-off-by: Firo Yang <firogm@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
John Ogness [Mon, 27 Apr 2015 11:52:33 +0000 (13:52 +0200)]
tty: serial: 8250: omap: synchronize rx_running
The rx_running flag should show if DMA is currently active. However
there is a window between when the flag is set/cleared and when
the DMA is started/stopped. Because the flag is queried from both
hard and soft irq contexts, the driver can make incorrect
decisions and do things like start a DMA transfer using a buffer
that is already setup to be used for a DMA transfer.
This patch adds a spinlock to synchronize the rx_running flag and
close the above mentioned window.
Signed-off-by: John Ogness <john.ogness@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
John Ogness [Mon, 27 Apr 2015 11:52:25 +0000 (13:52 +0200)]
dmaenegine: edma: allow pause/resume for non-cyclic mode
The 8250_omap serial driver relies on dmaengine_pause() actually
pausing the DMA transfer. Before this patch dmaengine_pause() is
a NOP for non-cylic DMA transfers. This allowed the 8250_omap
driver to read DMA buffers while the DMA was still active,
resulting in lost serial data.
Signed-off-by: John Ogness <john.ogness@linutronix.de> Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
John Ogness [Mon, 27 Apr 2015 11:52:15 +0000 (13:52 +0200)]
ARM: common: edma: clear completion interrupts on stop
When stopping a DMA transfer with interrupts disabled it is possible
that the DMA transfer completes before the events are cleared. In
this case the completion interrupt will be pending, causing a
completion callback after the transfer was stopped.
By clearing the completion interrupt for the stopping channel it is
ensured that no completion event will be generated after the stop.
Signed-off-by: John Ogness <john.ogness@linutronix.de> Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/built-in.o: In function `ifx_spi_free_device':
ifx6x60.c:(.text+0x96d9a): undefined reference to `dma_free_coherent'
drivers/built-in.o: In function `ifx_spi_spi_probe':
ifx6x60.c:(.text+0x978a2): undefined reference to `dma_alloc_coherent'
While DMA is optional in this driver, and is used only if
ifx_modem_platform_data.use_dma is set, there are currently no in-tree
users of ifx_modem_platform_data (and thus of this driver), so just make
it depend on HAS_DMA.
Qipan Li [Mon, 20 Apr 2015 08:10:23 +0000 (08:10 +0000)]
serial: sirf: move from clk_get to devm_clk_get
Signed-off-by: Qipan Li <Qipan.Li@csr.com> Signed-off-by: Barry Song <Baohua.Song@csr.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Qipan Li [Mon, 20 Apr 2015 08:10:22 +0000 (08:10 +0000)]
Revert "serial: sirf: add a new uart type support"
This reverts commit 52bec4ed4e("serial: sirf: add a new uart type
support").
we misunderstood the clock dependency in atlas7. Actually involved
several clocks are in a tree structure. we still only need to take
the leaf clock node for BT uarts.
Signed-off-by: Qipan Li <Qipan.Li@csr.com> Signed-off-by: Barry Song <Barry.Song@csr.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove CONFIG_SERIAL_DETECT_IRQ and CONFIG_SERIAL_MANY_PORTS, and
substitute all references to the proper 8250 Kconfig options. Now, the
actual Kconfig dependencies are not hidden when reading the code and
static analyzers are less confused.
The CONFIG_ prefix is reserved for Kconfig options in Make and CPP
syntax; static analysis tools rely on this convention. This patch
enforces this behavior for SERIAL_BFIN_{HARD_}CTSRTS.
Handle the changed flag of IRQ status in interruption
instead of handling it in tasklet due to the tasklet
may be scheduled more than once in one interruption.
Otherwise, the changed status may be processed more
than once which will lead to unexpected result. And
seriously, kernel will crash.
The 'CONFIG_' prefix is reserved for Kconfig options in Make and CPP
syntax only. Remove this prefix from the self-defined CPP identifier to
apply to this convention and make static analysis tools happy.
drivers/tty/serial/crisv10.c: remove dead #ifdef blocks
ETRAX_SERIAL_PROC_ENTRY is not defined in Kconfig. The affected #ifdef
block has not been compiled for years, and the embedded macro, PROCSTAT,
ended up as a NOOP. Hence, remove the block and all calls to PROCSTAT.
drivers/tty/serial/sh-sci.h: remove dead reference on ARCH_SH7372
The Kconfig option ARCH_SH7372 has been removed by commit 59b89af1d555
("ARM: shmobile: sh7372: Remove Legacy C SoC code"). This patch removes
the last reference on this Kconfig option.
Rasmus Villemoes [Tue, 31 Mar 2015 13:55:58 +0000 (15:55 +0200)]
tty: return tty->name directly from tty_name
All users of tty_name pass the return value (the provided buffer) to
some printf-like function. We can thus avoid the strcpy and, more
importantly, later remove the buf parameter completely, eliminating
the need for some 64 byte stack buffers.
Rasmus Villemoes [Tue, 31 Mar 2015 13:55:57 +0000 (15:55 +0200)]
tty: constify return type of tty_name
All users of tty_name pass the result directly to a printf-like
function. This means we can actually let tty_name return the literal
"NULL tty" or tty->name directly, avoiding the strcpy and a lot of
medium-sized stack buffers. In preparation for that, make the return
type const char*.
While at it, we can also constify the tty parameter.
Jiada Wang [Mon, 13 Apr 2015 09:31:43 +0000 (18:31 +0900)]
serial: imx: protect Soft Reset of port with lock
Previously Soft Reset (clear of SRST bit in UCR2 register)
of UART in startup is not protected by lock, which may have race
with console_write, as console_write may occur at anytime even
when UART port is shutdown.
To avoid this race, protect Soft reset of UART port with spin_lock.
Signed-off-by: Jiada Wang <jiada_wang@mentor.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Dave Martin [Mon, 27 Apr 2015 15:49:05 +0000 (16:49 +0100)]
serial/amba-pl011: Refactor and simplify TX FIFO handling
Commit 734745c serial/amba-pl011: Activate TX IRQ passively
adds some complexity and overhead in the form of a softirq
mechanism for transmitting in the absence of interrupts.
This patch simplifies the code flow to reduce the reliance on
subtle behaviour and avoid fragility under future maintenance.
To this end, the TX softirq mechanism is removed and instead
pl011_start_tx() will now simply stuff the FIFO until full
(guaranteeing future TX IRQs), or until there are no more chars
to write (in which case we don't care whether an IRQ happens).
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Jakub Kicinski <kubakici@wp.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jakub Kiciński observed that this patch can cause the pl011
driver to hang if if the only process with a pl011 port open is
killed by a signal, pl011_shutdown() can get called with an
arbitrary amount of data still in the FIFO.
Calling _shutdown() with the TX FIFO non-empty is questionable
behaviour and my itself be a bug.
Since the affected patch was speculative anyway, and brings limited
benefit, the simplest course is to remove the assumption that TXIS
will always be left asserted after the port is shut down.
Signed-off-by: Dave Martin <Dave.Martin@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Linus Torvalds [Mon, 4 May 2015 01:23:53 +0000 (18:23 -0700)]
Merge tag 'for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull ext4 fixes from Ted Ts'o:
"Some miscellaneous bug fixes and some final on-disk and ABI changes
for ext4 encryption which provide better security and performance"
* tag 'for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
ext4: fix growing of tiny filesystems
ext4: move check under lock scope to close a race.
ext4: fix data corruption caused by unwritten and delayed extents
ext4 crypto: remove duplicated encryption mode definitions
ext4 crypto: do not select from EXT4_FS_ENCRYPTION
ext4 crypto: add padding to filenames before encrypting
ext4 crypto: simplify and speed up filename encryption
Linus Torvalds [Mon, 4 May 2015 01:15:48 +0000 (18:15 -0700)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Pull drm fixes from Dave Airlie:
"One intel fix, one rockchip fix, and a bunch of radeon fixes for some
regressions from audio rework and vm stability"
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/i915/chv: Implement WaDisableShadowRegForCpd
drm/radeon: fix userptr return value checking (v2)
drm/radeon: check new address before removing old one
drm/radeon: reset BOs address after clearing it.
drm/radeon: fix lockup when BOs aren't part of the VM on release
drm/radeon: add SI DPM quirk for Sapphire R9 270 Dual-X 2G GDDR5
drm/radeon: adjust pll when audio is not enabled
drm/radeon: only enable audio streams if the monitor supports it
drm/radeon: only mark audio as connected if the monitor supports it (v3)
drm/radeon/audio: don't enable packets until the end
drm/radeon: drop dce6_dp_enable
drm/radeon: fix ordering of AVI packet setup
drm/radeon: Use drm_calloc_ab for CS relocs
drm/rockchip: fix error check when getting irq
MAINTAINERS: add entry for Rockchip drm drivers
Dave Airlie [Sun, 3 May 2015 22:56:47 +0000 (08:56 +1000)]
Merge tag 'drm-intel-fixes-2015-04-30' of git://anongit.freedesktop.org/drm-intel into drm-fixes
Just a single intel fix
* tag 'drm-intel-fixes-2015-04-30' of git://anongit.freedesktop.org/drm-intel:
drm/i915/chv: Implement WaDisableShadowRegForCpd
Dave Airlie [Sun, 3 May 2015 22:56:27 +0000 (08:56 +1000)]
Merge branch 'drm-next0420' of https://github.com/markyzq/kernel-drm-rockchip into drm-fixes
one fix and maintainers update
* 'drm-next0420' of https://github.com/markyzq/kernel-drm-rockchip:
drm/rockchip: fix error check when getting irq
MAINTAINERS: add entry for Rockchip drm drivers
Linus Torvalds [Sun, 3 May 2015 20:22:32 +0000 (13:22 -0700)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"This is three logical fixes (as 5 patches).
The 3ware class of drivers were causing an oops with multiqueue by
tearing down the command mappings after completing the command (where
the variables in the command used to tear down the mapping were
no-longer valid). There's also a fix for the qnap iscsi target which
was choking on us sending it commands that were too long and a fix for
the reworked aha1542 allocating GFP_KERNEL under a lock"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
3w-9xxx: fix command completion race
3w-xxxx: fix command completion race
3w-sas: fix command completion race
aha1542: Allocate memory before taking a lock
SCSI: add 1024 max sectors black list flag
Linus Torvalds [Sun, 3 May 2015 17:49:04 +0000 (10:49 -0700)]
Merge branch 'next' of git://git.infradead.org/users/vkoul/slave-dma
Pull slave dmaengine fixes from Vinod Koul:
"Here are the fixes in dmaengine subsystem for rc2:
- privatecnt fix for slave dma request API by Christopher
- warn fix for PM ifdef in usb-dmac by Geert
- fix hardware dependency for xgene by Jean"
* 'next' of git://git.infradead.org/users/vkoul/slave-dma:
dmaengine: increment privatecnt when using dma_get_any_slave_channel
dmaengine: xgene: Set hardware dependency
dmaengine: usb-dmac: Protect PM-only functions to kill warning
Linus Torvalds [Sun, 3 May 2015 17:28:36 +0000 (10:28 -0700)]
Merge tag 'powerpc-4.1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux
Pull powerpc fixes from Michael Ellerman:
- build fix for SMP=n in book3s_xics.c
- fix for Daniel's pci_controller_ops on powernv.
- revert the TM syscall abort patch for now.
- CPU affinity fix from Nathan.
- two EEH fixes from Gavin.
- fix for CR corruption from Sam.
- selftest build fix.
* tag 'powerpc-4.1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux:
powerpc/powernv: Restore non-volatile CRs after nap
powerpc/eeh: Delay probing EEH device during hotplug
powerpc/eeh: Fix race condition in pcibios_set_pcie_reset_state()
powerpc/pseries: Correct cpu affinity for dlpar added cpus
selftests/powerpc: Fix the pmu install rule
Revert "powerpc/tm: Abort syscalls in active transactions"
powerpc/powernv: Fix early pci_controller_ops loading.
powerpc/kvm: Fix SMP=n build error in book3s_xics.c
Jan Kara [Sun, 3 May 2015 03:58:32 +0000 (23:58 -0400)]
ext4: fix growing of tiny filesystems
The estimate of necessary transaction credits in ext4_flex_group_add()
is too pessimistic. It reserves credit for sb, resize inode, and resize
inode dindirect block for each group added in a flex group although they
are always the same block and thus it is enough to account them only
once. Also the number of modified GDT block is overestimated since we
fit EXT4_DESC_PER_BLOCK(sb) descriptors in one block.
Make the estimation more precise. That reduces number of requested
credits enough that we can grow 20 MB filesystem (which has 1 MB
journal, 79 reserved GDT blocks, and flex group size 16 by default).
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Davide Italiano [Sun, 3 May 2015 03:21:15 +0000 (23:21 -0400)]
ext4: move check under lock scope to close a race.
fallocate() checks that the file is extent-based and returns
EOPNOTSUPP in case is not. Other tasks can convert from and to
indirect and extent so it's safe to check only after grabbing
the inode mutex.
Signed-off-by: Davide Italiano <dccitaliano@gmail.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Cc: stable@vger.kernel.org
Lukas Czerner [Sun, 3 May 2015 01:36:55 +0000 (21:36 -0400)]
ext4: fix data corruption caused by unwritten and delayed extents
Currently it is possible to lose whole file system block worth of data
when we hit the specific interaction with unwritten and delayed extents
in status extent tree.
The problem is that when we insert delayed extent into extent status
tree the only way to get rid of it is when we write out delayed buffer.
However there is a limitation in the extent status tree implementation
so that when inserting unwritten extent should there be even a single
delayed block the whole unwritten extent would be marked as delayed.
At this point, there is no way to get rid of the delayed extents,
because there are no delayed buffers to write out. So when a we write
into said unwritten extent we will convert it to written, but it still
remains delayed.
When we try to write into that block later ext4_da_map_blocks() will set
the buffer new and delayed and map it to invalid block which causes
the rest of the block to be zeroed loosing already written data.
For now we can fix this by simply not allowing to set delayed status on
written extent in the extent status tree. Also add WARN_ON() to make
sure that we notice if this happens in the future.
This problem can be easily reproduced by running the following xfs_io.
This can be theoretically also reproduced by at random by running fsx,
but it's not very reliable, though on machines with bigger page size
(like ppc) this can be seen more often (especially xfstest generic/127)
Herbert Xu [Sat, 2 May 2015 14:29:19 +0000 (10:29 -0400)]
ext4 crypto: do not select from EXT4_FS_ENCRYPTION
This patch adds a tristate EXT4_ENCRYPTION to do the selections
for EXT4_FS_ENCRYPTION because selecting from a bool causes all
the selected options to be built-in, even if EXT4 itself is a
module.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Theodore Ts'o <tytso@mit.edu>
1) Receive packet length needs to be adjust by 2 on RX to accomodate
the two padding bytes in altera_tse driver. From Vlastimil Setka.
2) If rx frame is dropped due to out of memory in macb driver, we leave
the receive ring descriptors in an undefined state. From Punnaiah
Choudary Kalluri
3) Some netlink subsystems erroneously signal NLM_F_MULTI. That is
only for dumps. Fix from Nicolas Dichtel.
4) Fix mis-use of raw rt->rt_pmtu value in ipv4, one must always go via
the ipv4_mtu() helper. From Herbert Xu.
5) Fix null deref in bridge netfilter, and miscalculated lengths in
jump/goto nf_tables verdicts. From Florian Westphal.
6) Unhash ping sockets properly.
7) Software implementation of BPF divide did 64/32 rather than 64/64
bit divide. The JITs got it right. Fix from Alexei Starovoitov.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (30 commits)
ipv4: Missing sk_nulls_node_init() in ping_unhash().
net: fec: Fix RGMII-ID mode
net/mlx4_en: Schedule napi when RX buffers allocation fails
netxen_nic: use spin_[un]lock_bh around tx_clean_lock
net/mlx4_core: Fix unaligned accesses
mlx4_en: Use correct loop cursor in error path.
cxgb4: Fix MC1 memory offset calculation
bnx2x: Delay during kdump load
net: Fix Kernel Panic in bonding driver debugfs file: rlb_hash_table
net: dsa: Fix scope of eeprom-length property
net: macb: Fix race condition in driver when Rx frame is dropped
hv_netvsc: Fix a bug in netvsc_start_xmit()
altera_tse: Correct rx packet length
mlx4: Fix tx ring affinity_mask creation
tipc: fix problem with parallel link synchronization mechanism
tipc: remove wrong use of NLM_F_MULTI
bridge/nl: remove wrong use of NLM_F_MULTI
bridge/mdb: remove wrong use of NLM_F_MULTI
net: sched: act_connmark: don't zap skb->nfct
trivial: net: systemport: bcmsysport.h: fix 0x0x prefix
...
David S. Miller [Sat, 2 May 2015 02:02:47 +0000 (22:02 -0400)]
ipv4: Missing sk_nulls_node_init() in ping_unhash().
If we don't do that, then the poison value is left in the ->pprev
backlink.
This can cause crashes if we do a disconnect, followed by a connect().
Tested-by: Linus Torvalds <torvalds@linux-foundation.org> Reported-by: Wen Xu <hotdog3645@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
When we end I/O struct request with error, we need to pass
obj_request->length as @nr_bytes so that the entire obj_request worth
of bytes is completed. Otherwise block layer ends up confused and we
trip on
in rbd_img_obj_callback() due to more being true no matter what. We
already do it in most cases but we are missing some, in particular
those where we don't even get a chance to submit any obj_requests, due
to an early -ENOMEM for example.
A number of obj_request->xferred assignments seem to be redundant but
I haven't touched any of obj_request->xferred stuff to keep this small
and isolated.
Cc: Alex Elder <elder@linaro.org> Cc: stable@vger.kernel.org # 3.10+ Reported-by: Shawn Edwards <lesser.evil@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Theodore Ts'o [Fri, 1 May 2015 20:56:50 +0000 (16:56 -0400)]
ext4 crypto: add padding to filenames before encrypting
This obscures the length of the filenames, to decrease the amount of
information leakage. By default, we pad the filenames to the next 4
byte boundaries. This costs nothing, since the directory entries are
aligned to 4 byte boundaries anyway. Filenames can also be padded to
8, 16, or 32 bytes, which will consume more directory space.
Theodore Ts'o [Fri, 1 May 2015 20:56:45 +0000 (16:56 -0400)]
ext4 crypto: simplify and speed up filename encryption
Avoid using SHA-1 when calculating the user-visible filename when the
encryption key is available, and avoid decrypting lots of filenames
when searching for a directory entry in a directory block.
projects / karo-tx-linux.git / log