ioat2: fix performance regression
Subsystems (3)
PATCH 1/1 ioat2: fix performance regression
Commit 0793448 "DMAENGINE: generic channel status v2" changed the interface for how dma channel progress is retrieved. It inadvertently exported an internal helper function ioat_tx_status() instead of ioat_dma_tx_status(). The latter polls the hardware to get the latest completion state, while the helper just evaluates the current state without touching hardware. The effect is that we end up waiting for completion timeouts or descriptor allocation errors before the completion state is updated.
iperf (before fix):
[SUM] 0.0-41.3 sec 364 MBytes 73.9 Mbits/sec
iperf (after fix):
[SUM] 0.0- 4.5 sec 499 MBytes 940 Mbits/sec
This is a regression starting with 2.6.35.
---
Late breaking fix that can hopefully get into 2.6.36. I will queue a cleanup to clarify this confusing naming convention for 2.6.37.
Also available via git:
git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx.git fixes
Dan Williams (1):
ioat2: fix performance regression
drivers/dma/ioat/dma_v2.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/dma/ioat/dma_v2.c +1 -1
@@ -879,7 +879,7 @@ int __devinit ioat2_dma_probe(struct ioatdma_device *device, int dca) dma->device_issue_pending = ioat2_issue_pending; dma->device_alloc_chan_resources = ioat2_alloc_chan_resources; dma->device_free_chan_resources = ioat2_free_chan_resources; - dma->device_tx_status = ioat_tx_status; + dma->device_tx_status = ioat_dma_tx_status; err =ioat_probe (device);if (err)
Quoted 3 lines · Dan Williams
Commit 0793448 "DMAENGINE: generic channel status v2" changed the interface for how dma channel progress is retrieved. It inadvertently exported an internal helper function
ioat_tx_status()instead ofioat_dma_tx_status().
Yours,
Linus Walleij