On 18/10/2018 12:48, John Garry wrote:
On 18/10/2018 12:19, Robin Murphy wrote:
On 18/10/18 11:55, John Garry wrote:
[...]
@@ -976,18 +1019,19 @@ static int __arm_smmu_cmdq_issue_sync(struct
arm_smmu_device *smmu)
{
u64 cmd[CMDQ_ENT_DWORDS];
unsigned long flags;
- bool wfe = !!(smmu->features & ARM_SMMU_FEAT_SEV);
struct arm_smmu_cmdq_ent ent = { .opcode = CMDQ_OP_CMD_SYNC };
- int ret;
+ int sync_idx, sync_gen;
arm_smmu_cmdq_build_cmd(cmd, &ent);
spin_lock_irqsave(&smmu->cmdq.lock, flags);
- arm_smmu_cmdq_insert_cmd(smmu, cmd);
- ret = queue_poll_cons(&smmu->cmdq.q, true, wfe);
+ if (smmu->prev_cmd_opcode != CMDQ_OP_CMD_SYNC)
+ arm_smmu_cmdq_insert_cmd(smmu, cmd);
Hi Robin,
If we did stop rebuilding the non-MSI command as I suggested, then we
would not have the case of building the command and then discarding
it, right?
I suppose so. But that build/discard case can also be avoided by
applying patch 2/2 of this series ;)
So we can avoid the build/discard but the command is built under the
lock. Hopefully worth it.
However there is still scope (and - considering the locking - maybe more
need) for a static non-MSI CMD_SYNC command :)
Anyway, we'll look to get some numbers.
Cheers,
OK, I'll check it more detail. TBH, I found 2/2 hard to follow from just
the diff.
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
