On Tue, Oct 20, 2026 at 03:22:09PM +0000, Dawid Wesierski wrote: > Three fixes in the iavf traffic manager. > > Patch 1 removes a second free of the queue TC map. The fault is in main > only, after commit f6e33f09f766. No release has it. > > Patch 2 sets the queue ID and TC of each queue node in the queue > bandwidth message. Without it, a queue node with no shaper profile > resets queue 0, and its own queue keeps the rate limit that the PF > holds from an earlier process. > > Patch 3 returns -ENOTSUP from the TM ops that read qos_cap when the PF > does not grant VIRTCHNL_VF_OFFLOAD_QOS, for example the in-tree ice PF. > These ops cause a segmentation fault now. > > Tested with testpmd on an E830 VF, out-of-tree ice PF: > - Patch 1: three hierarchy commits on a started port. Without the > patch, an ASan build reports a heap-buffer-overflow in rte_free() > from iavf_hierarchy_commit() at the second commit. With it, there is > no report, also after port reset and close. > - Patch 2: one process sets 100 Mbit/s on two queues and exits. The > next process commits two queue nodes without a shaper profile. > Without the patch, only queue 0 gets back its full rate. With it, > both queues do. > - Patch 3: in-tree ice PF, so the VF has no QoS offload. Without the > patch, TM capabilities get, shaper profile add and node add give a > segmentation fault. With it, they return an error. > > v2: > - Patch 1: also remove the rte_free() of the map in iavf_uninit_vf() > (Bruce Richardson). > - Patch 2: commit message from Bruce Richardson. > - Add the Acked-by of Bruce Richardson to the three patches. > - Rebase on the current next-net-intel. > Series applied to dpdk-next-net-intel.
Thanks, /Bruce
