Hi Linus, please pull from:
git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx.git next
...to receive:
1/ Some reworks of the fsldma and coh dma drivers. The fsldma fixes
have been tested to enable async_tx operation chaining.
2/ A dma driver for the new mpc512x platform (base support for this
platform already merged through the powerpc tree).
3/ Interrupt coalescing and cleanups for the ioat driver
4/ Fixes for dmatest's RAID functionality
5/ Other random cleanups and fixes
All but the coh and ioat changes have spent time in -next. The coh
rework was recently broken up into a series of 5 patches. The ioat
changes have seen testing on an internal validation tree, but were
cherry-picked to this branch for release.
Thanks,
Dan
Anatolij Gustschin (1):
dmatest: correct raid6 PQ test
Dan Williams (8):
dmatest: fix handling of an even number of xor_sources
ioat3: cleanup, don't enable DCA completion writes
ioat3: use ioat2_quiesce()
ioat2: kill pending flag
ioat: close potential BUG_ON race in the descriptor cleanup path
ioat3: interrupt coalescing
ioat: cleanup ->timer_fn() and ->cleanup_fn() prototypes
Merge branch 'coh' into dmaengine
Ira Snyder (8):
fsldma: reduce kernel text size
fsldma: remove unused structure members
fsldma: rename struct fsl_dma_chan to struct fsldma_chan
fsldma: rename dest to dst for uniformity
fsldma: clean up the OF subsystem routines
fsldma: simplify IRQ probing and handling
fsldma: rename fsl_chan to chan
fsldma: major cleanups and fixes
Linus Walleij (6):
Debugging options for the DMA engine subsystem
DMAENGINE: COH 901 318 cleanups
DMAENGINE: COH 901 318 descriptor pool refactoring
DMAENGINE: COH 901 318 remove irq counting
DMAENGINE: COH 901 318 configure channel direction
DMAENGINE: COH 901 318 lli sg offset fix
Márton Németh (1):
dma: make Open Firmware device id constant
Piotr Ziecik (1):
dma: Add MPC512x DMA driver
Roel Kluin (2):
dma: cases IPU_PIX_FMT_BGRA32, BGR32 and ABGR32 are the same in ipu_ch_param_set_size()
iop-adma: redundant/wrong tests in iop_*_count()?
Steven J. Magnani (2):
fsldma: Fix cookie issues
fsldma: Fix cookie issues
Documentation/powerpc/dts-bindings/fsl/dma.txt | 8 +
arch/arm/include/asm/hardware/iop3xx-adma.h | 12 +-
arch/arm/mach-u300/include/mach/coh901318.h | 2 +-
drivers/dma/Kconfig | 23 +
drivers/dma/Makefile | 8 +
drivers/dma/coh901318.c | 182 ++--
drivers/dma/coh901318_lli.c | 23 +-
drivers/dma/dmatest.c | 8 +-
drivers/dma/fsldma.c | 1177 +++++++++++++-----------
drivers/dma/fsldma.h | 35 +-
drivers/dma/ioat/dma.c | 46 +-
drivers/dma/ioat/dma.h | 11 +-
drivers/dma/ioat/dma_v2.c | 70 +-
drivers/dma/ioat/dma_v2.h | 6 +-
drivers/dma/ioat/dma_v3.c | 64 +-
drivers/dma/ioat/registers.h | 2 +-
drivers/dma/ipu/ipu_idmac.c | 15 +-
drivers/dma/mpc512x_dma.c | 800 ++++++++++++++++
drivers/dma/ppc4xx/adma.c | 2 +-
include/linux/dmaengine.h | 2 +
20 files changed, 1717 insertions(+), 779 deletions(-)
create mode 100644 drivers/dma/mpc512x_dma.c
commit 272ca655090978bdaa2630fc44fb2c03da5576fd
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:33:59 2010 +0000
fsldma: reduce kernel text size
Some of the functions are written in a way where they use multiple reads
and writes where a single read/write pair could suffice. This shrinks the
kernel text size measurably, while making the functions easier to
understand.
add/remove: 0/0 grow/shrink: 1/4 up/down: 4/-196 (-192)
function old new delta
fsl_chan_set_request_count 120 124 +4
dma_halt 300 272 -28
fsl_chan_set_src_loop_size 208 156 -52
fsl_chan_set_dest_loop_size 208 156 -52
fsl_chan_xfer_ld_queue 500 436 -64
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 4ce0e953f6286777452bf07c83056342d6b9b257
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:00 2010 +0000
fsldma: remove unused structure members
Remove some unused members from the fsldma data structures. A few trivial
uses of struct resource were converted to use the stack rather than keeping
the memory allocated for the lifetime of the driver.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit a4f56d4b103d4e5d1a59a9118db0185a6bd1a83b
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:01 2010 +0000
fsldma: rename struct fsl_dma_chan to struct fsldma_chan
This is the beginning of a cleanup which will change all instances of
"fsl_dma" to "fsldma" to match the name of the driver itself.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 738f5f7e1ae876448cb7d9c82bea258b69386647
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:02 2010 +0000
fsldma: rename dest to dst for uniformity
Most functions in the standard library use "dst" as a parameter, rather
than "dest". This renames all use of "dest" to "dst" to match the usual
convention.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit e7a29151de1bd52081f27f149b68074fac0323be
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:03 2010 +0000
fsldma: clean up the OF subsystem routines
This fixes some errors in the cleanup paths of the OF subsystem, including
missing checks for ioremap failing. Also, some variables were renamed for
brevity.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit d3f620b2c4fecdc8e060b70e8d92d29fc01c6126
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:04 2010 +0000
fsldma: simplify IRQ probing and handling
The IRQ probing is needlessly complex. All off the 83xx device trees in
arch/powerpc/boot/dts/ specify 5 interrupts per DMA controller: one for the
controller, and one for each channel. These interrupts are all attached to
the same IRQ line.
This causes an interesting situation if two channels interrupt at the same
time. The per-controller handler will handle the first channel, and the
per-channel handler will handle the remaining channels.
Instead of this mess, we fix the bug in the per-controller handler, and
make it handle all channels that generated an interrupt. When a
per-controller handler is specified in the device tree, we prefer to use
the shared handler instead of the per-channel handler.
The 85xx/86xx controllers do not have a per-controller interrupt, and
instead use a per-channel interrupt. This behavior has not been changed.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit a1c03319018061304be28d131073ac13a5cb86fb
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:05 2010 +0000
fsldma: rename fsl_chan to chan
The name fsl_chan seems too long, so it has been shortened to chan. There
are only a few places where the higher level "struct dma_chan *chan" name
conflicts. These have been changed to "struct dma_chan *dchan" instead.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 9c3a50b7d7ec45da34e73cac66cde12dd6092dd8
Author: Ira Snyder <iws@xxxxxxxxxxxxxxxx>
Date: Wed Jan 6 13:34:06 2010 +0000
fsldma: major cleanups and fixes
Fix locking. Use two queues in the driver, one for pending transacions, and
one for transactions which are actually running on the hardware. Call
dma_run_dependencies() on descriptor cleanup so that the async_tx API works
correctly.
There are a number of places throughout the code where lists of descriptors
are freed in a loop. Create functions to handle this, and use them instead
of open-coding the loop each time.
Signed-off-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 4b1cf1facca31b7db2a61d8aa2ba40d5a93a0957
Author: Márton Németh <nm127@xxxxxxxxxxx>
Date: Tue Feb 2 23:41:06 2010 -0700
dma: make Open Firmware device id constant
The match_table field of the struct of_device_id is constant in <linux/of_platform.h>
so it is worth to make the initialization data also constant.
The semantic match that finds this kind of pattern is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@r@
disable decl_init,const_decl_init;
identifier I1, I2, x;
@@
struct I1 {
...
const struct I2 *x;
...
};
@s@
identifier r.I1, y;
identifier r.x, E;
@@
struct I1 y = {
.x = E,
};
@c@
identifier r.I2;
identifier s.E;
@@
const struct I2 E[] = ... ;
@depends on !c@
identifier r.I2;
identifier s.E;
@@
+ const
struct I2 E[] = ...;
// </smpl>
Signed-off-by: Márton Németh <nm127@xxxxxxxxxxx>
Cc: Julia Lawall <julia@xxxxxxx>
Cc: cocci@xxxxxxx
[dan.j.williams@xxxxxxxxx: resolved conflict with recent fsldma updates]
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 9ad7bd2944bd979ef4877cd439719be44c5f3b47
Author: Roel Kluin <roel.kluin@xxxxxxxxx>
Date: Wed Jan 20 01:25:56 2010 +0100
dma: cases IPU_PIX_FMT_BGRA32, BGR32 and ABGR32 are the same in ipu_ch_param_set_size()
In these cases the same statements are executed.
Signed-off-by: Roel Kluin <roel.kluin@xxxxxxxxx>
Acked-by: Guennadi Liakhovetski <g.liakhovetski@xxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 6ca3a7a96e91b1aa8c704153c992b191d35b5747
Author: Steven J. Magnani <steve@xxxxxxxxxxxxxxx>
Date: Thu Feb 25 13:39:30 2010 -0600
fsldma: Fix cookie issues
fsl_dma_tx_submit() only sets the cookie on the first descriptor of a
transaction. It should set the cookie on all.
Signed-off-by: Steven J. Magnani <steve@xxxxxxxxxxxxxxx>
Acked-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 76bd061f5c7b7550cdaed68ad6219ea7cee288fc
Author: Steven J. Magnani <steve@xxxxxxxxxxxxxxx>
Date: Sun Feb 28 22:18:16 2010 -0700
fsldma: Fix cookie issues
fsl_dma_update_completed_cookie() appears to calculate the last completed
cookie incorrectly in the corner case where DMA on cookie 1 is in progress
just following a cookie wrap.
Signed-off-by: Steven J. Magnani <steve@xxxxxxxxxxxxxxx>
Acked-by: Ira W. Snyder <iws@xxxxxxxxxxxxxxxx>
[dan.j.williams@xxxxxxxxx: fix an integer overflow warning with INT_MAX]
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 94de648d72c8bc833590523f22386d4babbea988
Author: Anatolij Gustschin <agust@xxxxxxx>
Date: Mon Feb 15 22:35:23 2010 +0100
dmatest: correct raid6 PQ test
The number of PQ sources specified by module parameter "pq_sources"
is always forced odd to fit into dmatest's destination verificaton
scheme. But number of PQ sources and coefficients as passed to the
driver's prep_dma_pq() is not adjusted accordingly.
Fix it now to get correct PQ testing results in the case passed
"pq_sources" parameter is even.
Signed-off-by: Anatolij Gustschin <agust@xxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Maciej Sosnowski <maciej.sosnowski@xxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 67b9124f734b22b30d9adf18c39fe795e2901070
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Sun Feb 28 22:20:18 2010 -0700
dmatest: fix handling of an even number of xor_sources
Just like commit ac5d73fc, we need to be careful to use 'src_cnt' as it
contains the fixed up number of xor sources (forced odd) to meet dmatest's
data verification scheme.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit f1acb878b6070941e844dfc4ca1b3b9e5a70426c
Author: Roel Kluin <roel.kluin@xxxxxxxxx>
Date: Sat Feb 13 15:54:03 2010 +0100
iop-adma: redundant/wrong tests in iop_*_count()?
When we reach the loop, len is at least 1, we only stay in the loop when
len is at least MAX_BYTE_COUNT + 1, MAX_BYTE_COUNT is subtracted in each
iteration. So when we leave the loop, or didn't take it, len is at least 1.
Testing whether len is non-zero appears redundant.
Signed-off-by: Roel Kluin <roel.kluin@xxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 6c664a8915f5341c2e7f1df0bb4b9b4a88f6ad77
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Feb 9 22:34:54 2010 +0100
Debugging options for the DMA engine subsystem
This adds Kconfig options for DEBUG and VERBOSE_DEBUG to the DMA
engine subsystem, I got tired of editing the Makefile manually
each time I want to debug things in here, modelled this on the
debug switches for other subsystems and works like a charm when
working on our DMA engines.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 0fb6f739bb612bc989d295056877374b749e721b
Author: Piotr Ziecik <kosmo@xxxxxxxxxxxx>
Date: Fri Feb 5 03:42:52 2010 +0000
dma: Add MPC512x DMA driver
Adds initial version of MPC512x DMA driver.
Only memory to memory transfers are currenly supported.
Signed-off-by: Piotr Ziecik <kosmo@xxxxxxxxxxxx>
Signed-off-by: Wolfgang Denk <wd@xxxxxxx>
Signed-off-by: Anatolij Gustschin <agust@xxxxxxx>
Cc: John Rigby <jcrigby@xxxxxxxxx>
Acked-by: Grant Likely <grant.likely@xxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 848ad121240f539e14a59eddd69e164aea9560b2
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Mar 2 14:17:15 2010 -0700
DMAENGINE: COH 901 318 cleanups
This cleans up the some debug code that was not working in the
COH 901 318 driver, adds some helpful comments and rearrange the
code a bit.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit b87108a772e001af3fa79f9cfd87b190375f47a2
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Mar 2 14:17:20 2010 -0700
DMAENGINE: COH 901 318 descriptor pool refactoring
This centralize some spread-out initialization of descriptors into
one function and cleans up the error paths.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 0b58828c923e57f1bfbbd2c4277ceb60666314fa
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Mar 2 14:17:44 2010 -0700
DMAENGINE: COH 901 318 remove irq counting
This removes the pointless irq counting for the COH 901 318, as
it turns out the hardware will only ever fire one IRQ for a linked
list anyway. In the process also a missing spinlock was introduced.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 516fd4305e5f5718475e81fe5c17c95888a8157b
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Mar 2 20:12:46 2010 +0100
DMAENGINE: COH 901 318 configure channel direction
This makes the COH 901 318 configure channel direction (to or from
device) dynamically, instead of being passed in from the platform
data. This was necessary in order to get the MMC/SD-card channel
bidirectional (all other channels on the U300 were either RX or
TX but this one was both). This also sets memcpy() alignent to
even 2^2 (32bit) boundaries, which makes the memcpy() stress tests
start working.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 56a5d3cf21c71963c8fc506e9b9d3f71641d9c71
Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Date: Tue Mar 2 20:12:56 2010 +0100
DMAENGINE: COH 901 318 lli sg offset fix
This makes the COH 901 318 respect the scatter offset field by using
the sg_phys() rather than the sg_dma_address() so we get a pointer
to the actual data we want to send rather than the beginning of the
buffer. Also initialize the lli:s a bit more thoroughly.
Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 773d9e2d8dbf02cfaf65786cf9100eef02c9fda4
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 11:47:42 2010 -0700
ioat3: cleanup, don't enable DCA completion writes
We already disallow raid operations while DCA is globally enabled, so
having it locally enabled is a nop and confusing when reading the code.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit b372ec2d900a5b50e47ef9e9624536ad146236be
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 11:47:42 2010 -0700
ioat3: use ioat2_quiesce()
Replace open coded ioat2_quiesce() call in ioat3_restart_channel
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit 281befa5592b0c5f9a3856b5666c62ac66d3d9ee
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 11:47:43 2010 -0700
ioat2: kill pending flag
The pending == 2 case no longer exists in the driver so, we can use
ioat2_ring_pending() outside the lock to determine if there might be any
descriptors in the ring that the hardware has not seen.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit aa75db0080603bae27961c0502812dfd0f522bb3
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 21:21:10 2010 -0700
ioat: close potential BUG_ON race in the descriptor cleanup path
Since ioat_cleanup_preamble() and the update of the last completed
descriptor are not synchronized there is a chance that two cleanup threads
can see descriptors to clean. If the first cleans up all pending
descriptors then the second will trigger the BUG_ON.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit b9cc98697d1ca35a86bbb708acc6d93993c28f0f
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 21:21:13 2010 -0700
ioat3: interrupt coalescing
The hardware automatically disables further interrupts after each event
until rearmed. This allows a delay to be injected between the occurence
of the interrupt and the running of the cleanup routine. The delay is
scaled by the descriptor backlog and then written to the INTRDELAY
register which specifies the number of microseconds to hold off
interrupt delivery after an interrupt event occurs. According to
powertop this reduces the interrupt rate from ~5000 intr/s to ~150
intr/s per without affecting throughput (simple dd to a raid6 array).
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit aa4d72ae946a4fa40486b871717778734184fa29
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 21:21:13 2010 -0700
ioat: cleanup ->timer_fn() and ->cleanup_fn() prototypes
If the calling convention of ->timer_fn() and ->cleanup_fn() are unified
across hardware versions we can drop parameters to ioat_init_channel() and
unify ioat_is_dma_complete() implementations.
Both ->timer_fn() and ->cleanup_fn() are modified to expect a struct
dma_chan pointer.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
commit dd58ffcf5a5352fc10820c8ffbcd5fed416a2c3a
Merge: aa4d72a 56a5d3c
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Mar 3 21:22:21 2010 -0700
Merge branch 'coh' into dmaengine
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
