New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

vchiq_2835_arm: Implement a DMA pool for small bulk transfers #2699

Merged
merged 4 commits into from Oct 2, 2018

Conversation

Projects
None yet
3 participants
@detule
Contributor

detule commented Oct 2, 2018

An attempt to address #2680 by using a DMA pool for small allocations during bulk transfers.

Thanks to @pelwell and @lategoodbye for reviewing V1 of this patch.

detule and others added some commits Oct 1, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

@pelwell pelwell merged commit e24b1f6 into raspberrypi:rpi-4.14.y Oct 2, 2018

@pelwell

This comment has been minimized.

Contributor

pelwell commented Oct 2, 2018

Thank you!

@popcornmix popcornmix referenced this pull request Oct 2, 2018

Closed

cma_alloc: alloc failed #2680

popcornmix added a commit that referenced this pull request Oct 2, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 2, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

nathanchance added a commit to nathanchance/pi-kernel that referenced this pull request Oct 2, 2018

Merge remote-tracking branch 'upstream/rpi-4.14.y' into rpi-4.14.y
* upstream/rpi-4.14.y:
  vchiq_2835_arm: Implement a DMA pool for small bulk transfers (raspberrypi#2699)
  Revert "ASoC: cs4265: SOC_SINGLE register value error fix"

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>

Conflicts:
	sound/soc/codecs/cs4265.c

pelwell referenced this pull request in detule/linux Oct 3, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

popcornmix added a commit that referenced this pull request Oct 8, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 8, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 15, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 15, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 22, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Oct 22, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 4, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 4, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>
@lategoodbye

This comment has been minimized.

Contributor

lategoodbye commented Nov 4, 2018

@detule Are you interested in upstreaming your patch? In case you are not familiar with the procedure you can contact me: stefan.wahren@i2se.com

@detule

This comment has been minimized.

Contributor

detule commented Nov 5, 2018

@lategoodbye

Thanks for reaching out - definitely interested; will contact you.

popcornmix added a commit that referenced this pull request Nov 12, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 12, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 12, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 13, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 15, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>

popcornmix added a commit that referenced this pull request Nov 15, 2018

vchiq_2835_arm: Implement a DMA pool for small bulk transfers (#2699)
During a bulk transfer we request a DMA allocation to hold the
scatter-gather list.  Most of the time, this allocation is small
(<< PAGE_SIZE), however it can be requested at a high enough frequency
to cause fragmentation and/or stress the CMA allocator (think time
spent in compaction here, or during allocations elsewhere).

Implement a pool to serve up small DMA allocations, falling back
to a coherent allocation if the request is greater than
VCHIQ_DMA_POOL_SIZE.

Signed-off-by: Oliver Gjoneski <ogjoneski@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment