UCT/MLX5: Fix dp_ordering logic - needed to disable DDP when no AR #10489
+52
−40
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tvegas1
reviewed
Feb 13, 2025
| ucs_config_sprintf_ternary_auto(ddp_enable_str, sizeof(ddp_enable_str), | ||
| &config->ddp_enable, NULL); | ||
| ucs_error("%s/%s: cannot set ar_enable=%s ddp_enable=%s cap=%s " | ||
| "supp_force=%d (min=%s max=%s force=%d)", |
brminich
reviewed
Feb 14, 2025
Comment on lines
+633
to
+634
| min_dp_ordering = ucs_max(min_dp_ordering, | ||
| UCT_IB_MLX5_DP_ORDERING_OOO_ALL); |
There was a problem hiding this comment.
why do we need ucs_max here? can just assign to UCT_IB_MLX5_DP_ORDERING_OOO_ALL
There was a problem hiding this comment.
it's true that since we calculate min/max here right after they are initialized to the initial values, no need to really use min.max, but IMO this improves code readability and shows that applying DDP and AR configurations is interchangeable.
Comment on lines
+649
to
+650
| max_dp_ordering = ucs_min(max_dp_ordering, | ||
| UCT_IB_MLX5_DP_ORDERING_IBTA); |
There was a problem hiding this comment.
why do we need ucs_min here?
| } else if (config->ddp_enable == UCS_NO) { | ||
| /* No more than OOO_RW would be used */ | ||
| max_dp_ordering = ucs_min(max_dp_ordering, | ||
| UCT_IB_MLX5_DP_ORDERING_OOO_RW); |
There was a problem hiding this comment.
maybe initialize force=1 and set it to 0 just here?
There was a problem hiding this comment.
Fixed to logic to calculate "force", so the test would also not fail when AR_ENABLE=n and AR is not supported(and cannot be forced)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why
Fix a failure to initialize RC/DC iface with DDP_ENABLE=n, when the NIC does not support OOO/AR.
Happens because of an issue in the current logic: DDP_ENABLE=n implies we want MD to support UCT_IB_MLX5_MD_FLAG_DP_ORDERING_FORCE (presumable to force-disable), while in fact it is not needed.
How
Examine DDP and AR configurations independently; fail if they conflict; pick the maxinal level that is both supported and needed.
Failure example: