In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
gve: fix header buffer corruption with header-split and HW-GRO<\/p>\n
The DQO RX datapath programs a per-buffer-queue-descriptor
\nheader_buf_addr at post time and reads the split header back at
\ncompletion time. Both the post and the read currently index the
\nheader buffer by queue position rather than by the buffer’s identity:<\/p>\n
– post (gve_rx_post_buffers_dqo): header_buf_addr is computed from
\n bufq->tail
\n – read (gve_rx_dqo): the header is read from desc_idx (the completion
\n queue head index)<\/p>\n
This relies on the buffer-queue index and the completion-queue index
\nbeing equal for the start of every packet, i.e. on the device consuming
\nposted buffers and returning completions in the exact same order. That
\nassumption does not hold once HW-GRO is enabled with multiple
\nflows: coalesced segments are accepted and completed in an order that
\nmay differ from the order buffers were posted, and segments from
\ndifferent flows may interleave.<\/p>\n
That results in two problems:<\/p>\n
1. Wrong header slot on read. Because the read offset is derived from
\n the completion index (desc_idx) while the device wrote the header to
\n the address programmed for the buffer’s buf_id, the driver can copy
\n a header belonging to a different packet. This shows up as
\n throughput drop (about 30% drop and large numbers of TCP
\n retransmissions) with header-split and HW-GRO both enabled and many
\n streams.<\/p>\n
2. Header buffer reused while still owned by the device. The driver
\n advances bufq->head by one per completion and re-posts buffers based
\n on that. Arrival of N RX completions only guarantees that at least N
\n RX buffer descriptors have been read by the device. It does not
\n guarantee that the device has relinquished the ownership of all the
\n buffers corresponding to those N descriptors. With out-of-order
\n completions (e.g. the completion for a packet copied into buffer N
\n arrives before the completion for a packet copied into buffer N-1),
\n the driver can re-post and overwrite a header buffer that the device
\n is still going to write into, corrupting the header of a packet
\n whose completion has not yet been processed.<\/p>\n
Fix both issues by indexing the header buffer by buf_id on both the post
\nand read paths. Reading from buf_id’s slot is therefore always correct
\nregardless of completion ordering (fixes problem 1).<\/p>\n
Indexing by buf_id also ties each header slot to the lifetime of its
\nbuffer state. A buffer state is only returned to the free\/recycle lists
\nwhen its own completion (buf_id) is processed, so its header slot can
\nonly be re-posted after the device is done with it. This makes header
\nslot reuse safe under out-of-order completions (fixes problem 2).<\/p>\n
Allocate (gve_rx_alloc_hdr_bufs) and free (gve_rx_free_hdr_bufs) the
\nheader buffers based on num_buf_states to match the buf_id indexing.<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
gve: fix header buffer corruption with header-split and HW-GRO<\/p>\n
The DQO RX datapath programs a per-buffer-queue-descriptor
\nheader_b<\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-6745","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/6745","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=6745"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6745"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6745"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}