"value":"In the Linux kernel, the following vulnerability has been resolved:\n\ncifs: Fix server re-repick on subrequest retry\n\nWhen a subrequest is marked for needing retry, netfs will call\ncifs_prepare_write() which will make cifs repick the server for the op\nbefore renegotiating credits; it then calls cifs_issue_write() which\ninvokes smb2_async_writev() - which re-repicks the server.\n\nIf a different server is then selected, this causes the increment of\nserver->in_flight to happen against one record and the decrement to happen\nagainst another, leading to misaccounting.\n\nFix this by just removing the repick code in smb2_async_writev(). As this\nis only called from netfslib-driven code, cifs_prepare_write() should\nalways have been called first, and so server should never be NULL and the\npreparatory step is repeated in the event that we do a retry.\n\nThe problem manifests as a warning looking something like:\n\n WARNING: CPU: 4 PID: 72896 at fs/smb/client/smb2ops.c:97 smb2_add_credits+0x3f0/0x9e0 [cifs]\n ...\n RIP: 0010:smb2_add_credits+0x3f0/0x9e0 [cifs]\n ...\n smb2_writev_callback+0x334/0x560 [cifs]\n cifs_demultiplex_thread+0x77a/0x11b0 [cifs]\n kthread+0x187/0x1d0\n ret_from_fork+0x34/0x60\n ret_from_fork_asm+0x1a/0x30\n\nWhich may be triggered by a number of different xfstests running against an\nAzure server in multichannel mode. generic/249 seems the most repeatable,\nbut generic/215, generic/249 and generic/308 may also show it."