CVE-2025-40027

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net/9p: fix double req put in p9_fd_cancelled<br /> <br /> Syzkaller reports a KASAN issue as below:<br /> <br /> general protection fault, probably for non-canonical address 0xfbd59c0000000021: 0000 [#1] PREEMPT SMP KASAN NOPTI<br /> KASAN: maybe wild-memory-access in range [0xdead000000000108-0xdead00000000010f]<br /> CPU: 0 PID: 5083 Comm: syz-executor.2 Not tainted 6.1.134-syzkaller-00037-g855bd1d7d838 #0<br /> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014<br /> RIP: 0010:__list_del include/linux/list.h:114 [inline]<br /> RIP: 0010:__list_del_entry include/linux/list.h:137 [inline]<br /> RIP: 0010:list_del include/linux/list.h:148 [inline]<br /> RIP: 0010:p9_fd_cancelled+0xe9/0x200 net/9p/trans_fd.c:734<br /> <br /> Call Trace:<br /> <br /> p9_client_flush+0x351/0x440 net/9p/client.c:614<br /> p9_client_rpc+0xb6b/0xc70 net/9p/client.c:734<br /> p9_client_version net/9p/client.c:920 [inline]<br /> p9_client_create+0xb51/0x1240 net/9p/client.c:1027<br /> v9fs_session_init+0x1f0/0x18f0 fs/9p/v9fs.c:408<br /> v9fs_mount+0xba/0xcb0 fs/9p/vfs_super.c:126<br /> legacy_get_tree+0x108/0x220 fs/fs_context.c:632<br /> vfs_get_tree+0x8e/0x300 fs/super.c:1573<br /> do_new_mount fs/namespace.c:3056 [inline]<br /> path_mount+0x6a6/0x1e90 fs/namespace.c:3386<br /> do_mount fs/namespace.c:3399 [inline]<br /> __do_sys_mount fs/namespace.c:3607 [inline]<br /> __se_sys_mount fs/namespace.c:3584 [inline]<br /> __x64_sys_mount+0x283/0x300 fs/namespace.c:3584<br /> do_syscall_x64 arch/x86/entry/common.c:51 [inline]<br /> do_syscall_64+0x35/0x80 arch/x86/entry/common.c:81<br /> entry_SYSCALL_64_after_hwframe+0x6e/0xd8<br /> <br /> This happens because of a race condition between:<br /> <br /> - The 9p client sending an invalid flush request and later cleaning it up;<br /> - The 9p client in p9_read_work() canceled all pending requests.<br /> <br /> Thread 1 Thread 2<br /> ...<br /> p9_client_create()<br /> ...<br /> p9_fd_create()<br /> ...<br /> p9_conn_create()<br /> ...<br /> // start Thread 2<br /> INIT_WORK(&amp;m-&gt;rq, p9_read_work);<br /> p9_read_work()<br /> ...<br /> p9_client_rpc()<br /> ...<br /> ...<br /> p9_conn_cancel()<br /> ...<br /> spin_lock(&amp;m-&gt;req_lock);<br /> ...<br /> p9_fd_cancelled()<br /> ...<br /> ...<br /> spin_unlock(&amp;m-&gt;req_lock);<br /> // status rewrite<br /> p9_client_cb(m-&gt;client, req, REQ_STATUS_ERROR)<br /> // first remove<br /> list_del(&amp;req-&gt;req_list);<br /> ...<br /> <br /> spin_lock(&amp;m-&gt;req_lock)<br /> ...<br /> // second remove<br /> list_del(&amp;req-&gt;req_list);<br /> spin_unlock(&amp;m-&gt;req_lock)<br /> ...<br /> <br /> Commit 74d6a5d56629 ("9p/trans_fd: Fix concurrency del of req_list in<br /> p9_fd_cancelled/p9_read_work") fixes a concurrency issue in the 9p filesystem<br /> client where the req_list could be deleted simultaneously by both<br /> p9_read_work and p9_fd_cancelled functions, but for the case where req-&gt;status<br /> equals REQ_STATUS_RCVD.<br /> <br /> Update the check for req-&gt;status in p9_fd_cancelled to skip processing not<br /> just received requests, but anything that is not SENT, as whatever<br /> changed the state from SENT also removed the request from its list.<br /> <br /> Found by Linux Verification Center (linuxtesting.org) with Syzkaller.<br /> <br /> [updated the check from status == RECV || status == ERROR to status != SENT]