[git pull] dmaengine, async_tx, and raid6 for 2.6.32
From: Dan Williams
Date: Thu Sep 17 2009 - 13:25:37 EST
Hi Neil, please pull from:
git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx.git next
...to receive:
1/ Async_tx extensions for raid6 including syndrome generation and
recovery. A test suite module, raid6test based on hpa's userspace test,
has been added to unit test the recovery code.
2/ md/raid6 conversion to the new async routines. This includes a
percpu conversion of stripe cache operations.
3/ An experimental multicore option for raid processing. This is a
trivial (i.e. easily revertible) extension on top of the percpu
conversion. It is marked experimental as it currently appears to
oversubscribe the cpus, but still shows some modest gains to throughput
(+4% seq write for 16-way SATA raid6).
4/ A new dma driver for SuperH platforms (shdma)
5/ Support for version 3.2 Intel(R) QuickData Technology devices which
adds raid5/6 offload support to the ioatdma driver. This work also
includes a re-factoring of the driver [1].
6/ The fsldma driver picks up "slave" support.
7/ Various dmaengine fix ups
The bulk of this has been in linux-next for at least a week. Patches to
address feedback from you and Maciej are copied below. I chose to
append these rather than rebase to preserve the commit numbers as seen
by linux-next.
Thanks,
Dan
Atsushi Nemoto (1):
dmaengine: Move all map_sg/unmap_sg for slave channel to its client
Dan Williams (81):
Merge branch 'dmaengine' into async-tx-raid6
async_tx: rename zero_sum to val
async_tx: kill ASYNC_TX_DEP_ACK flag
async_tx: structify submission arguments, add scribble
async_xor: permit callers to pass in a 'dma/page scribble' region
md/raid6: release spare page at ->stop()
ioat: move to drivers/dma/ioat/
md/raid6: move the spare page to a percpu allocation
md/raid5,6: add percpu scribble region for buffer lists
async_tx: add sum check flags
async_tx: kill needless module_{init|exit}
async_tx: remove walk of tx->parent chain in dma_wait_for_async_tx
async_tx: add support for asynchronous GF multiplication
async_tx: add support for asynchronous RAID6 recovery operations
dmatest: add pq support
async_tx: raid6 recovery self test
iop-adma: cleanup iop_adma_run_tx_complete_actions
iop-adma: fix lockdep false positive
iop-adma: P+Q support for iop13xx adma engines
iop-adma: P+Q self test
md/raid5: factor out mark_uptodate from ops_complete_compute5
md/raid6: asynchronous raid6 operations
md/raid6: asynchronous handle_parity_check6
md/raid456: distribute raid processing over multiple cores
Merge commit 'v2.6.31-rc1' into dmaengine
ioat: move definitions to dma.h
ioat: convert ioat_probe to pcim/devm
ioat: cleanup some long deref chains and 80 column collisions
ioat: kill function prototype ifdef guards
ioat: split ioat_dma_probe into core/version-specific routines
ioat: fix type mismatch for ->dmacount
ioat: define descriptor control bit-field
ioat1: move descriptor allocation from submit to prep
ioat: fix self test interrupts
ioat: prepare the code for ioat[12]_dma_chan split
ioat2,3: convert to a true ring buffer
ioat1: kill unused unmap parameters
ioat: add some dev_dbg() calls
ioat: cleanup completion status reads
ioat: ignore reserved bits for chancnt and xfercap
ioat: preserve chanctrl bits when re-arming interrupts
ioat: ___devinit annotate the initialization paths
ioat1: trim ioat_dma_desc_sw
ioat: switch watchdog and reset handler from workqueue to timer
ioat2,3: dynamically resize descriptor ring
net_dma: poll for a descriptor after allocation failure
Merge branch 'md-raid6-accel' into ioat3.2
dmaengine: add fence support
dmaengine, async_tx: add a "no channel switch" allocator
dmaengine: cleanup unused transaction types
dmaengine, async_tx: support alignment checks
ioat2+: add fence support
ioat3: hardware version 3.2 register / descriptor definitions
ioat3: split ioat3 support to its own file, add memset
ioat: add 'ioat' sysfs attributes
ioat3: enable dca for completion writes
ioat3: xor support
ioat3: xor self test
ioat3: pq support
ioat3: support xor via pq descriptors
ioat3: interrupt descriptor support
ioat3: segregate raid engines
Merge branch 'ioat' into dmaengine
dw_dmac: implement a private tx_list
fsldma: implement a private tx_list
iop-adma: implement a private tx_list
ioat: implement a private tx_list
mv_xor: implement a private tx_list
at_hdmac: implement a private tx_list
txx9dmac: implement a private tx_list
dmaengine: kill tx_list
ioat2,3: cacheline align software descriptor allocations
Merge branch 'iop-raid6' into async-tx-next
Merge branch 'dmaengine' into async-tx-next
Merge commit 'md/for-linus' into async-tx-next
async_tx: remove HIGHMEM64G restriction
ioat: driver version 4.0
md/raid6: eliminate BUG_ON with side effect
md/raid6: cleanup ops_run_compute6_2
ioat2: clarify ring size limits
raid6test: fix stack overflow
Ira Snyder (2):
fsldma: split apart external pause and request count features
fsldma: Add DMA_SLAVE support
Maciej Sosnowski (1):
dca: registering requesters in multiple dca domains
Nobuhiro Iwamatsu (1):
dmaengine: sh: Add Support SuperH DMA Engine driver
Roland Dreier (2):
Add MODULE_DEVICE_TABLE() so ioatdma module is autoloaded
I/OAT: Convert to PCI_VDEVICE()
Stephen Hemminger (1):
dca: module load should not be an error message
Tom Picard (1):
ioat3: ioat3.2 pci ids for Jasper Forest
Yuri Tikhonov (5):
md/raid5,6: common schedule_reconstruction for raid5/6
md/raid6: asynchronous handle_stripe_fill6
md/raid6: asynchronous handle_stripe_dirtying6
md/raid6: asynchronous handle_stripe6
md/raid6: remove synchronous infrastructure
Documentation/crypto/async-tx-api.txt | 75 +-
arch/arm/include/asm/hardware/iop3xx-adma.h | 81 +-
arch/arm/include/asm/hardware/iop_adma.h | 3 +
arch/arm/mach-iop13xx/include/mach/adma.h | 119 ++-
arch/arm/mach-iop13xx/setup.c | 17 +-
arch/arm/plat-iop/adma.c | 4 +-
arch/powerpc/include/asm/fsldma.h | 136 ++
arch/sh/drivers/dma/Kconfig | 12 +-
arch/sh/drivers/dma/Makefile | 3 +-
arch/sh/include/asm/dma-sh.h | 13 +
crypto/async_tx/Kconfig | 9 +
crypto/async_tx/Makefile | 3 +
crypto/async_tx/async_memcpy.c | 44 +-
crypto/async_tx/async_memset.c | 43 +-
crypto/async_tx/async_pq.c | 395 +++++
crypto/async_tx/async_raid6_recov.c | 455 +++++
crypto/async_tx/async_tx.c | 87 +-
crypto/async_tx/async_xor.c | 207 ++--
crypto/async_tx/raid6test.c | 240 +++
drivers/dca/dca-core.c | 124 ++-
drivers/dma/Kconfig | 14 +-
drivers/dma/Makefile | 4 +-
drivers/dma/at_hdmac.c | 60 +-
drivers/dma/at_hdmac_regs.h | 1 +
drivers/dma/dmaengine.c | 94 +-
drivers/dma/dmatest.c | 40 +
drivers/dma/dw_dmac.c | 50 +-
drivers/dma/dw_dmac_regs.h | 1 +
drivers/dma/fsldma.c | 288 +++-
drivers/dma/fsldma.h | 4 +-
drivers/dma/ioat.c | 202 ---
drivers/dma/ioat/Makefile | 2 +
drivers/dma/{ioat_dca.c => ioat/dca.c} | 13 +-
drivers/dma/ioat/dma.c | 1238 ++++++++++++++
drivers/dma/ioat/dma.h | 337 ++++
drivers/dma/ioat/dma_v2.c | 870 ++++++++++
drivers/dma/ioat/dma_v2.h | 190 +++
drivers/dma/ioat/dma_v3.c | 1220 ++++++++++++++
drivers/dma/ioat/hw.h | 215 +++
drivers/dma/ioat/pci.c | 210 +++
.../dma/{ioatdma_registers.h => ioat/registers.h} | 54 +-
drivers/dma/ioat_dma.c | 1741 --------------------
drivers/dma/ioatdma.h | 165 --
drivers/dma/ioatdma_hw.h | 70 -
drivers/dma/iop-adma.c | 491 +++++-
drivers/dma/iovlock.c | 10 +
drivers/dma/mv_xor.c | 7 +-
drivers/dma/mv_xor.h | 4 +-
drivers/dma/shdma.c | 786 +++++++++
drivers/dma/shdma.h | 64 +
drivers/dma/txx9dmac.c | 24 +-
drivers/dma/txx9dmac.h | 1 +
drivers/idle/i7300_idle.c | 20 +-
drivers/md/Kconfig | 26 +
drivers/md/raid5.c | 1475 ++++++++++-------
drivers/md/raid5.h | 28 +-
drivers/mmc/host/atmel-mci.c | 9 +-
include/linux/async_tx.h | 129 ++-
include/linux/dca.h | 11 +-
include/linux/dmaengine.h | 179 ++-
include/linux/pci_ids.h | 10 +
61 files changed, 9119 insertions(+), 3308 deletions(-)
create mode 100644 arch/powerpc/include/asm/fsldma.h
create mode 100644 crypto/async_tx/async_pq.c
create mode 100644 crypto/async_tx/async_raid6_recov.c
create mode 100644 crypto/async_tx/raid6test.c
delete mode 100644 drivers/dma/ioat.c
create mode 100644 drivers/dma/ioat/Makefile
rename drivers/dma/{ioat_dca.c => ioat/dca.c} (98%)
create mode 100644 drivers/dma/ioat/dma.c
create mode 100644 drivers/dma/ioat/dma.h
create mode 100644 drivers/dma/ioat/dma_v2.c
create mode 100644 drivers/dma/ioat/dma_v2.h
create mode 100644 drivers/dma/ioat/dma_v3.c
create mode 100644 drivers/dma/ioat/hw.h
create mode 100644 drivers/dma/ioat/pci.c
rename drivers/dma/{ioatdma_registers.h => ioat/registers.h} (84%)
delete mode 100644 drivers/dma/ioat_dma.c
delete mode 100644 drivers/dma/ioatdma.h
delete mode 100644 drivers/dma/ioatdma_hw.h
create mode 100644 drivers/dma/shdma.c
create mode 100644 drivers/dma/shdma.h
Late arrivals:
commit 1b6df6930994d5d027375b07ac9da63644eb5758
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Sep 16 21:03:29 2009 -0700
raid6test: fix stack overflow
Testing on x86_64 with NDISKS=255 yields:
do_IRQ: modprobe near stack overflow (cur:ffff88007d19c000,sp:ffff88007d19c128)
...and eventually
general protection fault: 0000 [#1]
Moving the scribble buffers off the stack allows the test to complete
successfully.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
diff --git a/crypto/async_tx/raid6test.c b/crypto/async_tx/raid6test.c
index 98c83ca..3ec27c7 100644
--- a/crypto/async_tx/raid6test.c
+++ b/crypto/async_tx/raid6test.c
@@ -28,6 +28,7 @@
#define NDISKS 16 /* Including P and Q */
static struct page *dataptrs[NDISKS];
+static addr_conv_t addr_conv[NDISKS];
static struct page *data[NDISKS+3];
static struct page *spare;
static struct page *recovi;
@@ -69,7 +70,6 @@ static char disk_type(int d, int disks)
static void raid6_dual_recov(int disks, size_t bytes, int faila, int failb, struct page **ptrs)
{
struct async_submit_ctl submit;
- addr_conv_t addr_conv[disks];
struct completion cmp;
struct dma_async_tx_descriptor *tx = NULL;
enum sum_check_flags result = ~0;
@@ -156,7 +156,6 @@ static int test_disks(int i, int j, int disks)
static int test(int disks, int *tests)
{
- addr_conv_t addr_conv[disks];
struct dma_async_tx_descriptor *tx;
struct async_submit_ctl submit;
struct completion cmp;
commit 376ec37667b510453f5a62fcd95d762786e6a0a9
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Sep 16 15:16:50 2009 -0700
ioat2: clarify ring size limits
With the addition of ioat_max_alloc_order it is not clear what the
maximum allocation order is, so document that in the modinfo. Also take
an opportunity to kill a stray semicolon.
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@xxxxxxxxx>
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
diff --git a/drivers/dma/ioat/dma_v2.c b/drivers/dma/ioat/dma_v2.c
index 5d6ac49..8fd0b59 100644
--- a/drivers/dma/ioat/dma_v2.c
+++ b/drivers/dma/ioat/dma_v2.c
@@ -42,18 +42,19 @@
int ioat_ring_alloc_order = 8;
module_param(ioat_ring_alloc_order, int, 0644);
MODULE_PARM_DESC(ioat_ring_alloc_order,
- "ioat2+: allocate 2^n descriptors per channel (default: n=8)");
+ "ioat2+: allocate 2^n descriptors per channel"
+ " (default: 8 max: 16)");
static int ioat_ring_max_alloc_order = IOAT_MAX_ORDER;
module_param(ioat_ring_max_alloc_order, int, 0644);
MODULE_PARM_DESC(ioat_ring_max_alloc_order,
- "ioat2+: upper limit for dynamic ring resizing (default: n=16)");
+ "ioat2+: upper limit for ring size (default: 16)");
void __ioat2_issue_pending(struct ioat2_dma_chan *ioat)
{
void * __iomem reg_base = ioat->base.reg_base;
ioat->pending = 0;
- ioat->dmacount += ioat2_ring_pending(ioat);;
+ ioat->dmacount += ioat2_ring_pending(ioat);
ioat->issued = ioat->head;
/* make descriptor updates globally visible before notifying channel */
wmb();
commit 6c910a78e495b4c1778a8b136b37fe3c05712730
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Sep 16 12:24:54 2009 -0700
md/raid6: cleanup ops_run_compute6_2
Neil says:
"It is correct as it stands, but the fact that every branch in
the 'if' part ends with a 'return' isn't immediately obvious,
so it is clearer if we are explicit about the if / then / else
structure."
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 0a5f03d..1898eda 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -810,7 +810,7 @@ ops_run_compute6_2(struct stripe_head *sh, struct raid5_percpu *percpu)
BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
BUG_ON(!test_bit(R5_Wantcompute, &tgt2->flags));
- /* we need to open-code set_syndrome_sources to handle to the
+ /* we need to open-code set_syndrome_sources to handle the
* slot number conversion for 'faila' and 'failb'
*/
for (i = 0; i < disks ; i++)
@@ -879,18 +879,21 @@ ops_run_compute6_2(struct stripe_head *sh, struct raid5_percpu *percpu)
return async_gen_syndrome(blocks, 0, count+2,
STRIPE_SIZE, &submit);
}
- }
-
- init_async_submit(&submit, ASYNC_TX_FENCE, NULL, ops_complete_compute,
- sh, to_addr_conv(sh, percpu));
- if (failb == syndrome_disks) {
- /* We're missing D+P. */
- return async_raid6_datap_recov(syndrome_disks+2, STRIPE_SIZE,
- faila, blocks, &submit);
} else {
- /* We're missing D+D. */
- return async_raid6_2data_recov(syndrome_disks+2, STRIPE_SIZE,
- faila, failb, blocks, &submit);
+ init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
+ ops_complete_compute, sh,
+ to_addr_conv(sh, percpu));
+ if (failb == syndrome_disks) {
+ /* We're missing D+P. */
+ return async_raid6_datap_recov(syndrome_disks+2,
+ STRIPE_SIZE, faila,
+ blocks, &submit);
+ } else {
+ /* We're missing D+D. */
+ return async_raid6_2data_recov(syndrome_disks+2,
+ STRIPE_SIZE, faila, failb,
+ blocks, &submit);
+ }
}
}
commit 2d6e4ecc87d20299bcb249dd62efbd73496744c3
Author: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Wed Sep 16 12:11:54 2009 -0700
md/raid6: eliminate BUG_ON with side effect
As pointed out by Neil it should be possible to build a driver with all
BUG_ON statements deleted. It's bad form to have a BUG_ON with a side
effect.
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 9b00a22..0a5f03d 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -3214,8 +3214,10 @@ static bool handle_stripe6(struct stripe_head *sh)
/* now count some things */
if (test_bit(R5_LOCKED, &dev->flags)) s.locked++;
if (test_bit(R5_UPTODATE, &dev->flags)) s.uptodate++;
- if (test_bit(R5_Wantcompute, &dev->flags))
- BUG_ON(++s.compute > 2);
+ if (test_bit(R5_Wantcompute, &dev->flags)) {
+ s.compute++;
+ BUG_ON(s.compute > 2);
+ }
if (test_bit(R5_Wantfill, &dev->flags)) {
s.to_fill++;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/