In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
bnxt_en: Prevent queue stop with deferred completions<\/p>\n
When the driver receives a burst of packets, it can mark a BD with the
\nNO_CMPL bit to defer completions. The expectation is that the last
\npacket in the ring will have this bit unset and the completion generated
\nby that packet will cleanup that packet and the ones preceding it. This
\nhelps to reduce the number of completions fired.<\/p>\n
The suppressed completions are controlled by the driver and the number
\nof packets with suppressed completions scales with the size of the ring.
\nSW USO packets, on the other hand, have an upper bound on the maximum
\nnumber of BDs which can be consumed which does not scale with the ring
\nsize.<\/p>\n
So, for small rings it is possible that: a burst of packets is handed to
\nthe driver, the driver defers completions for all of the packets because
\nthe number of free descriptors stays above the threshold in the driver.
\nThen, a USO packet arrives, but the number of BDs available is not
\nenough and the USO code exits early.<\/p>\n
In this case, you end up in a state where the ring is full of packets
\nwith their completions suppressed, which can cause the queue to stop and
\nnever be restarted.<\/p>\n
Assuming default CONFIG_MAX_SKB_FRAGS, this is only possible for small
\nrings (<= 457 descriptors, below the driver default value) when
\na burst of packets fills the ring, followed by a large USO packet that
\ncan't fit. For larger rings, the delta between the completion
\nsuppression threshold and the BDs required for SW USO is large enough
\nthat completions will fire and this case is unreachable.<\/p>\n
This issue was pointed out by Sashiko and while it seems fairly unlikely
\ngiven that the queue size must be small to trigger this, it is indeed
\npossible.<\/p>\n
Fix this by tracking the last BD which deferred completions and
\ncentralizing the logic for deciding when to ring the doorbell. The NO_CMPL
\nbit is now cleared in bnxt_txr_db_kick(), so every doorbell site is
\ncovered, including the SW USO early exit. This guarantees the ring always
\nends in a BD which generates a completion to clean it and wake the queue.<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
bnxt_en: Prevent queue stop with deferred completions<\/p>\n
When the driver receives a burst of packets, it can mark a BD with the
\nNO_CM<\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-29001","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/29001","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln"}],"about":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/types\/vuln"}],"wp:attachment":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=29001"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=29001"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=29001"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}