Support for Multiple Firmwares by JaiOCP · Pull Request #2331 · opencomputeproject/SAI

JaiOCP · 2026-08-13T21:55:21Z

Modern switches have embedded controller to run firmware for low latency applications.
This PR bring in support for installing and monitoring firmware based on the hardware support.

Older PR #2319 is closed

azure-pipelines · 2026-08-13T21:55:31Z

Azure Pipelines: There may be pipelines that require an authorized user to comment /azp run to run.

JaiOCP · 2026-08-13T21:57:47Z

@kcudnik @AnandhiDhanabalan @crameshk @eladnachman @gananthateerta @eladnachman @tjchadaga

Please approve this PR as the old one is closed.
All the review comments are addressed there.
#2319

kcudnik · 2026-08-14T12:28:44Z

why you are closing colder PR and creating new ones ? i already review older one and now i need to revierw again this one, i saw you do it already multiple times, this makes review proces lot harder

JaiOCP · 2026-08-17T18:33:03Z

why you are closing colder PR and creating new ones ? i already review older one and now i need to revierw again this one, i saw you do it already multiple times, this makes review proces lot harder

Kamil,
PR gets diverged so much that no new rebase or squash works. It infact makes commit history very messy with lots of new merges showing up in the audit record.

JaiOCP · 2026-08-17T19:27:52Z

@AnandhiDhanabalan Can you please approve the PR urgently? We need to close on this for next release.

tjchadaga · 2026-08-17T23:16:37Z

why you are closing colder PR and creating new ones ? i already review older one and now i need to revierw again this one, i saw you do it already multiple times, this makes review proces lot harder

Kamil, PR gets diverged so much that no new rebase or squash works. It infact makes commit history very messy with lots of new merges showing up in the audit record.

Jai, you should be able to squash commit, rebase and fix this issue. Please reach out before you close the PRs, I can try to fix it for you

AnandhiDhanabalan

@JaiOCP , thanks for addressing the comments. I noticed a few from the previous PR are still pending, so I’ve re-added them here. Could you please check?

tjchadaga · 2026-08-31T17:55:58Z

@JaiOCP, could you please address pending comments and update the branch?

author JaiOCP <jai.kumar@broadcom.com> 1786658065 -0700 committer JaiOCP <jai.kumar@broadcom.com> 1788461916 -0700 parent 9f91525 author JaiOCP <jai.kumar@broadcom.com> 1786658065 -0700 committer JaiOCP <jai.kumar@broadcom.com> 1788461347 -0700 Support for Multiple Firmwares Signed-off-by: JaiOCP <jai.kumar@broadcom.com> Add wide label attribute for objects with a label attribute (opencomputeproject#2327) Summary: This is similar to opencomputeproject#1158, opencomputeproject#1407, opencomputeproject#1430, opencomputeproject#2322. SAI label attributes are of type char, which maps to sai_attribute_value_t::chardata[32]. That is 32 bytes including the NUL terminator, so 31 usable characters. That can be too small for real identifiers. Add a LABEL_WIDE attribute of type sai_s8_list_t alongside each existing char label attribute. A list keeps the label length open, so the NOS can choose the length it needs instead of being constrained by an arbitrary limit in the SAI spec. Each new attribute is appended at the end of its attribute enum so that no existing attribute is renumbered. Objects covered: counter, ACL counter, LAG, mirror session, prefix compression table, policer, scheduler, next hop group, virtual router, UDF group and router interface. The existing char attributes are marked deprecated. Exactly one of the two attributes may be set to a non-default value; setting both is invalid. Signed-off-by: Shrikrishna (Shri) Khare <skhare@meta.com> Signed-off-by: Jasmeet Bagga <jasmeetbagga@meta.com> Signed-off-by: Ron He <zecheng@meta.com> Co-authored-by: Ron He <zecheng@meta.com> NHG Unequal Cost Members (opencomputeproject#2330) Signed-off-by: JaiOCP <jai.kumar@broadcom.com> Link Down Debounce (opencomputeproject#2284) Link Debounce Signed-off-by: JaiOCP <jai.kumar@broadcom.com> [debugcounter] Add support for SAI_IN_DROP_REASON_EGRESS_LINK_DOWN (opencomputeproject#2333) Extend the `sai_in_drop_reason_t` enum with `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` to support tracking packets dropped at ingress when the target egress port/interface is down. While `SAI_IN_DROP_REASON_L3_EGRESS_LINK_DOWN` specifically targets routed (L3) packets, certain ASIC architectures and pipelines drop traffic at ingress whenever the destination egress link is down regardless of packet type, or do not distinguish between L2 and L3 drops for this condition. Adding `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` enables debug counters to generically capture egress link down drops across both L2 (bridged) and L3 (routed) traffic. Change-Id: Ifb3b440f79f0eacccb6138b4d02410dcb0d02ebd Signed-off-by: Tommy Smail <tommysmail@google.com> fix NHG attribute comment (opencomputeproject#2317) Signed-off-by: Chikkegowda Chikkaiah <cchikkai@cisco.com> SAI header major version update to v1.19.0 (opencomputeproject#2342) Signed-off-by: Tejaswini Chadaga <tchadaga@microsoft.com> Support Multiple FW Signed-off-by: JaiOCP <jai.kumar@broadcom.com> NHG Unequal Cost Members (opencomputeproject#2330) Signed-off-by: JaiOCP <jai.kumar@broadcom.com> [debugcounter] Add support for SAI_IN_DROP_REASON_EGRESS_LINK_DOWN (opencomputeproject#2333) Extend the `sai_in_drop_reason_t` enum with `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` to support tracking packets dropped at ingress when the target egress port/interface is down. While `SAI_IN_DROP_REASON_L3_EGRESS_LINK_DOWN` specifically targets routed (L3) packets, certain ASIC architectures and pipelines drop traffic at ingress whenever the destination egress link is down regardless of packet type, or do not distinguish between L2 and L3 drops for this condition. Adding `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` enables debug counters to generically capture egress link down drops across both L2 (bridged) and L3 (routed) traffic. Change-Id: Ifb3b440f79f0eacccb6138b4d02410dcb0d02ebd Signed-off-by: Tommy Smail <tommysmail@google.com> fix NHG attribute comment (opencomputeproject#2317) Signed-off-by: Chikkegowda Chikkaiah <cchikkai@cisco.com> SAI header major version update to v1.19.0 (opencomputeproject#2342) Signed-off-by: Tejaswini Chadaga <tchadaga@microsoft.com> Support for Multiple FW Signed-off-by: JaiOCP <jai.kumar@broadcom.com> NHG Unequal Cost Members (opencomputeproject#2330) Signed-off-by: JaiOCP <jai.kumar@broadcom.com> [debugcounter] Add support for SAI_IN_DROP_REASON_EGRESS_LINK_DOWN (opencomputeproject#2333) Extend the `sai_in_drop_reason_t` enum with `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` to support tracking packets dropped at ingress when the target egress port/interface is down. While `SAI_IN_DROP_REASON_L3_EGRESS_LINK_DOWN` specifically targets routed (L3) packets, certain ASIC architectures and pipelines drop traffic at ingress whenever the destination egress link is down regardless of packet type, or do not distinguish between L2 and L3 drops for this condition. Adding `SAI_IN_DROP_REASON_EGRESS_LINK_DOWN` enables debug counters to generically capture egress link down drops across both L2 (bridged) and L3 (routed) traffic. Change-Id: Ifb3b440f79f0eacccb6138b4d02410dcb0d02ebd Signed-off-by: Tommy Smail <tommysmail@google.com> fix NHG attribute comment (opencomputeproject#2317) Signed-off-by: Chikkegowda Chikkaiah <cchikkai@cisco.com> SAI header major version update to v1.19.0 (opencomputeproject#2342) Signed-off-by: Tejaswini Chadaga <tchadaga@microsoft.com> ACL match route destination (opencomputeproject#2329) * Add support for ACL destination match Signed-off-by: Jason Bos <jbos@cisco.com> Support for Multiple FW Signed-off-by: JaiOCP <jai.kumar@broadcom.com>

kcudnik · 2026-09-04T10:32:02Z

    CHECK_STRUCT_SIZE(sai_acl_resource_t, 12);
    CHECK_STRUCT_SIZE(sai_attr_capability_t, 3);
-    CHECK_STRUCT_SIZE(sai_attribute_t, 48);
+    CHECK_STRUCT_SIZE(sai_attribute_t, 64);


this breaks backward compatybility

kcudnik · 2026-09-04T10:32:40Z

/azp run

azure-pipelines · 2026-09-04T10:32:49Z

Azure Pipelines: Successfully started running 1 pipeline(s).

JaiOCP mentioned this pull request Aug 13, 2026

Support for running firmware in embedded controllers #2319

Closed

eladnachman approved these changes Aug 16, 2026

View reviewed changes

AnandhiDhanabalan reviewed Aug 18, 2026

View reviewed changes

Comment thread inc/saifw.h

Comment thread inc/saifw.h

Comment thread doc/fw/SAI-FW-API.md

tjchadaga added the reviewed PR is discussed in SAI Meeting label Aug 26, 2026

JaiOCP force-pushed the FW branch 2 times, most recently from 4a74d02 to ca06a9f Compare September 3, 2026 18:51

JaiOCP force-pushed the FW branch from 78c95bd to 3524051 Compare September 3, 2026 19:00

Merge branch 'master' into FW

7254ce7

kcudnik reviewed Sep 4, 2026

View reviewed changes

kcudnik approved these changes Sep 4, 2026

View reviewed changes

Navigation Menu

Uh oh!

Support for Multiple Firmwares - #2331

Support for Multiple Firmwares#2331
JaiOCP wants to merge 2 commits into
opencomputeproject:masterfrom
JaiOCP:FW

JaiOCP commented Aug 13, 2026

Uh oh!

azure-pipelines Bot commented Aug 13, 2026

Uh oh!

JaiOCP commented Aug 13, 2026

Uh oh!

kcudnik commented Aug 14, 2026

Uh oh!

JaiOCP commented Aug 17, 2026

Uh oh!

JaiOCP commented Aug 17, 2026

Uh oh!

tjchadaga commented Aug 17, 2026

Uh oh!

AnandhiDhanabalan left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tjchadaga commented Aug 31, 2026

Uh oh!

kcudnik Sep 4, 2026

Uh oh!

kcudnik commented Sep 4, 2026

Uh oh!

azure-pipelines Bot commented Sep 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Uh oh!

Conversation

JaiOCP commented Aug 13, 2026

Uh oh!

azure-pipelines Bot commented Aug 13, 2026

Uh oh!

JaiOCP commented Aug 13, 2026

Uh oh!

kcudnik commented Aug 14, 2026

Uh oh!

JaiOCP commented Aug 17, 2026

Uh oh!

JaiOCP commented Aug 17, 2026

Uh oh!

tjchadaga commented Aug 17, 2026

Uh oh!

AnandhiDhanabalan left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

tjchadaga commented Aug 31, 2026

Uh oh!

kcudnik Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

kcudnik commented Sep 4, 2026

Uh oh!

azure-pipelines Bot commented Sep 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants