{"id":14942,"date":"2026-09-04T19:02:19","date_gmt":"2026-09-04T11:02:19","guid":{"rendered":"https:\/\/www.54iter.cn\/?vuln=cve-2026-63888"},"modified":"2026-09-04T19:02:19","modified_gmt":"2026-09-04T11:02:19","slug":"cve-2026-63888","status":"publish","type":"vuln","link":"https:\/\/www.54iter.cn\/?vuln=cve-2026-63888","title":{"rendered":"CVE-2026-63888 [CRITICAL] &#8211; In the Linux kernel, the following vulnerability has been resolved:\n\nscsi: target: iscsi:"},"content":{"rendered":"<p>In the Linux kernel, the following vulnerability has been resolved:<\/p>\n<p>scsi: target: iscsi: Fix CRC overread and double-free in iscsit_handle_text_cmd()<\/p>\n<p>Two latent bugs in the Text-phase handler, both present since the<br \/>\noriginal LIO integration in commit e48354ce078c (&#8220;iscsi-target: Add<br \/>\niSCSI fabric support for target v4.1&#8221;):<\/p>\n<p>1) DataDigest CRC buffer overread (4 bytes past text_in).<\/p>\n<p>   text_in is kzalloc()&#8217;d at ALIGN(payload_length, 4).  rx_size is then<br \/>\n   incremented by ISCSI_CRC_LEN to make room for the received DataDigest<br \/>\n   in the iovec, but the same (now-bumped) rx_size is passed as the<br \/>\n   buffer length to iscsit_crc_buf():<\/p>\n<p>       if (conn-&gt;conn_ops-&gt;DataDigest) {<br \/>\n               &#8230;<br \/>\n               rx_size += ISCSI_CRC_LEN;<br \/>\n       }<br \/>\n       &#8230;<br \/>\n       if (conn-&gt;conn_ops-&gt;DataDigest) {<br \/>\n               data_crc = iscsit_crc_buf(text_in, rx_size, 0, NULL);<\/p>\n<p>   iscsit_crc_buf() walks rx_size bytes of text_in with crc32c(), so<br \/>\n   when DataDigest is negotiated it reads 4 bytes past the end of the<br \/>\n   text_in allocation.  KASAN reproduces this directly on the unpatched<br \/>\n   mainline tree as slab-out-of-bounds in crc32c() called from the Text<br \/>\n   PDU path.  The OOB bytes feed crc32c() and are then compared against<br \/>\n   the initiator-supplied checksum, so the value does not flow back to<br \/>\n   the attacker, but the kernel does read past the buffer on every Text<br \/>\n   PDU with DataDigest=CRC32C.<\/p>\n<p>   Fix by passing the actual padded payload length<br \/>\n   (ALIGN(payload_length, 4)) that was used for the kzalloc().<\/p>\n<p>2) Stale cmd-&gt;text_in_ptr re-free (double-free) on ERL&gt;0 bad DataDigest<br \/>\n   drop.<\/p>\n<p>   On DataDigest mismatch with ErrorRecoveryLevel &gt; 0 the handler<br \/>\n   silently drops the PDU and lets the initiator plug the CmdSN gap:<\/p>\n<p>               kfree(text_in);<br \/>\n               return 0;<\/p>\n<p>   cmd-&gt;text_in_ptr still points at the freed buffer.  The next Text<br \/>\n   Request on the same ITT re-enters iscsit_setup_text_cmd(), which<br \/>\n   unconditionally does<\/p>\n<p>       kfree(cmd-&gt;text_in_ptr);<br \/>\n       cmd-&gt;text_in_ptr = NULL;<\/p>\n<p>   freeing the same pointer a second time.  Session teardown via<br \/>\n   iscsit_release_cmd() has the same shape and hits the same double-free<br \/>\n   if the connection is dropped before a second Text Request arrives.<\/p>\n<p>   On an unmodified mainline tree the bug-1 CRC overread fires first on<br \/>\n   the initial valid Text Request and perturbs the subsequent state, so<br \/>\n   #4 was isolated by building a kernel with only the bug-1 hunk of this<br \/>\n   patch applied plus temporary printk() observability around the three<br \/>\n   relevant kfree() sites.  The observability prints are not part of<br \/>\n   this patch.  On that build, a three-PDU Text Request sequence after<br \/>\n   login produces two back-to-back splats:<\/p>\n<p>       BUG: KASAN: double-free in iscsit_setup_text_cmd+0x??<br \/>\n       BUG: KASAN: double-free in iscsit_release_cmd+0x??<\/p>\n<p>   showing the same pointer freed in the ERL&gt;0 drop path and again in<br \/>\n   iscsit_setup_text_cmd() (next Text Request on the same ITT) and once<br \/>\n   more in iscsit_release_cmd() (session teardown).  On distro kernels<br \/>\n   with CONFIG_SLAB_FREELIST_HARDENED=y (default) the double-free<br \/>\n   becomes a remote kernel BUG(); on non-hardened kernels it corrupts<br \/>\n   the slab freelist.<\/p>\n<p>   Fix by clearing cmd-&gt;text_in_ptr after the kfree() in the ERL&gt;0 drop<br \/>\n   path.  With both hunks applied #4 is directly observable on the stock<br \/>\n   tree without observability printks; fixing bug-1 alone would mask #4<br \/>\n   less, not more, so the hunks are submitted together.<\/p>\n<p>Both fixes are one-liners.  The Text PDU state machine is unchanged and<br \/>\nthe wire protocol is unaffected.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>In the Linux kernel, the following vulnerability has been resolved:<\/p>\n<p>scsi: target: iscsi: Fix CRC overread and double-free in iscsit_handle_text_cmd()<\/p>\n<p>Two latent bugs in the Text-phase handler, both <\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-14942","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/14942","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=14942"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=14942"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=14942"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}