Re: [PATCH v9 00/12] iommu/arm-smmu-v3: Implement Runtime/System Sleep ops
From: Jason Gunthorpe <jgg@nvidia.com>
Date: 2026-08-25 20:15:03
Also in:
linux-iommu
On Tue, Aug 25, 2026 at 06:50:12PM +0000, Pranjal Shrivastava wrote:
On Tue, Aug 25, 2026 at 10:33:42AM -0300, Jason Gunthorpe wrote:quoted
On Tue, Jul 28, 2026 at 09:09:16PM +0000, Pranjal Shrivastava wrote:quoted
[v9] - Replaced atomic_fetch_or_relaxed() with atomic_fetch_or() for correct ordering while asserting the STOP_FLAG - Moved arm_smmu_drain_queues() after the cmdq->lock wait loop in runtime_suspend() as per feedback in v8. - Refactored suspend sequence to use arm_smmu_disable_irqs() and synchronize_irq() for safe GERROR handling. - Fixed an off-by-one error in suspend polling timeouts. - Dropped arm_smmu_cmdq_can_elide() check from inv_range - Renamed arm_smmu_cmdq_can_elide() to arm_smmu_is_active(). - Removed redundant bitwise masking from owner_prod reads. - Added comments to explain locking, gating and elision better. - Added comments to explain secondary cmdq draining and gating. - Rebased on arm/smmu/updatesThere are alot of scary looking sashiko remarks, are you planning to resend this with some of them addressed?Yes, I was first hoping to discuss them on the list! Thanks a lot for initiating that discussion :) [BTW, I'm interested to know if you used some tool to pull in sashiko feedback in your replies, such a tool seems helpful!]
I have been testing 'b4 review tui' to help my workflow. I have too
many patches to review and I'm trying to move them all into b4 and out
of my scattered mail boxes.
If you setup sashiko.dev integration:
[b4]
sashiko-url = https://sashiko.dev
Then when you load a series it just automagically shows the sashiko
stuff (press f to expand them) and you can easially quote them when
replying.
It can be a bit rough to ge there though, you have to get a series
applied to make it to that screen, so eg, the whitespace damage here
made that a pain.
However.. I had AI make an AI instruction for another AI to use AI to
to automagically fix git am conflicts to get *something* so I can use
the tool..
But I like that the patches end up applied on a check out branch and I
can poke around with my editor too..
Jason