In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
net: skbuff: do not leave stale header offsets after pskb_carve()<\/p>\n
pskb_carve_inside_header() and pskb_carve_inside_nonlinear() remove
\nthe first bytes of a packet and reallocate skb->head.<\/p>\n
All the headers that were present before the operation are gone,
\nbut both functions call skb_headers_offset_update(skb, 0), which
\nis a no-op : skb->mac_header, skb->network_header,
\nskb->transport_header and skb->csum_start keep their old values and
\nnow describe bytes which are no longer there.<\/p>\n
Both helpers size the new head from the old skb_end_offset(), so the
\nstale offsets still land inside the new allocation. They point past
\nskb_tail_pointer() though, to bytes that were never initialized.<\/p>\n
pskb_carve_inside_nonlinear() is the worst case, because it leaves a
\nzombie skb with an empty linear part (skb->data ==
\nskb_tail_pointer(skb), skb_headlen(skb) == 0), while
\nskb_mac_header_was_set() is still true and skb->mac_header is way
\nahead of skb->data.<\/p>\n
The only user of pskb_extract() is rds_tcp_data_recv(), and the
\ncarved skb is queued on tinc->ti_skb_list. When the RDS incoming
\nmessage is released, rds_tcp_inc_free() calls skb_queue_purge(),
\nwhich frees the skbs with SKB_DROP_REASON_QUEUE_PURGE. This is
\nvisible from drop_monitor, which then tries to pull back to the
\n(bogus) mac header :<\/p>\n
skbuff: __skb_pull(len=234)
\nskb len=6968 data_len=6968 headroom=0 headlen=0 tailroom=0
\nend-tail=384 mac=(234,14) mac_len=14 net=(248,40) trans=288
\nshinfo(txflags=0 nr_frags=1 gso(size=1428 type=16 segs=5))
\ncsum(0x100120 start=288 offset=16 ip_summed=3 complete_sw=0 valid=1 level=0)
\nhash(0x7b446c6c sw=0 l4=1) proto=0x86dd pkttype=0 iif=60
\nkernel BUG at .\/include\/linux\/skbuff.h:2847!<\/p>\n
Add skb_carve_reset_headers() to mark the mac and transport headers
\nas not set, reset the network header, clear skb->mac_len, and drop
\na now meaningless CHECKSUM_PARTIAL (csum_start no longer describes
\nanything).<\/p>\n
Invalidate the inner offsets as well. Unlike mac_header and
\ntransport_header they have no “unset” sentinel, so a leftover
\nnon-zero value still looks like a real header. Zero
\nskb->inner_mac_header, skb->inner_network_header,
\nskb->inner_transport_header, skb->inner_protocol and
\nskb->encapsulation, so that all the header state is invalidated in
\none place.<\/p>\n
v2: fixed an inaccurate changelog. The stale offsets stay inside the
\n new skb->head, which is never smaller than the old one, they
\n simply point past skb_tail_pointer() to bytes that are gone.
\n Thanks to Xuanqiang Luo for insisting on this.
\n Also invalidate the inner header state, as suggested by the
\n netdev AI review :
\n https:\/\/netdev-ai.bots.linux.dev\/sashiko\/#\/patchset\/20260911114922.621937-1-edumazet%40google.com<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
net: skbuff: do not leave stale header offsets after pskb_carve()<\/p>\n
pskb_carve_inside_header() and pskb_carve_inside_nonlinear() rem<\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-33340","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/33340","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=33340"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=33340"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=33340"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}