CVE-2025-39948

Description

In the Linux kernel, the following vulnerability has been resolved:ice: fix Rx page leak on multi-buffer framesThe ice_put_rx_mbuf() function handles calling ice_put_rx_buf() for eachbuffer in the current frame. This function was introduced as part ofhandling multi-buffer XDP support in the ice driver.It works by iterating over the buffers from first_desc up to 1 plus thetotal number of fragments in the frame, cached from before the XDP programwas executed.If the hardware posts a descriptor with a size of 0, the logic used inice_put_rx_mbuf() breaks. Such descriptors get skipped and dont get addedas fragments in ice_add_xdp_frag. Since the buffer isnt counted as afragment, we do not iterate over it in ice_put_rx_mbuf(), and thus we dontcall ice_put_rx_buf().Because we dont call ice_put_rx_buf(), we dont attempt to re-use thepage or free it. This leaves a stale page in the ring, as we dontincrement next_to_alloc.The ice_reuse_rx_page() assumes that the next_to_alloc has been incrementedproperly, and that it always points to a buffer with a null page. Sincethis function doesnt check, it will happily recycle a page over the topof the next_to_alloc buffer, losing track of the old page.Note that this leak only occurs for multi-buffer frames. Theice_put_rx_mbuf() function always handles at least one buffer, so asingle-buffer frame will always get handled correctly. It is not clearprecisely why the hardware hands us descriptors with a size of 0 sometimes,but it happens somewhat regularly with jumbo frames used by 9K MTU.To fix ice_put_rx_mbuf(), we need to make sure to call ice_put_rx_buf() onall buffers between first_desc and next_to_clean. Borrow the logic of asimilar function in i40e used for this same purpose. Use the same logicalso in ice_get_pgcnts().Instead of iterating over just the number of fragments, use a loop whichiterates until the current index reaches to the next_to_clean element justpast the current frame. Unlike i40e, the ice_put_rx_mbuf() function doescall ice_put_rx_buf() on the last buffer of the frame indicating the end ofpacket.For non-linear (multi-buffer) frames, we need to take care when adjustingthe pagecnt_bias. An XDP program might release fragments from the tail ofthe frame, in which case that fragment page is already released. Onlyupdate the pagecnt_bias for the first descriptor and fragments stillremaining post-XDP program. Take care to only access the shared info forfragmented buffers, as this avoids a significant cache miss.The xdp_xmit value only needs to be updated if an XDP program is run, andonly once per packet. Drop the xdp_xmit pointer argument fromice_put_rx_mbuf(). Instead, set xdp_xmit in the ice_clean_rx_irq() functiondirectly. This avoids needing to pass the argument and avoids an extrabit-wise OR for each buffer in the frame.Move the increment of the ntc local variable to ensure its updated *before*all calls to ice_get_pgcnts() or ice_put_rx_mbuf(), as the loop logicrequires the index of the element just after the current frame.Now that we use an index pointer in the ring to identify the packet, we nolonger need to track or cache the number of fragments in the rx_ring.

Risk Information

Associated Vulnerability

Patch Details

References

https://nvd.nist.gov/vuln/detail/CVE-2023-1234
https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2023-1234