CVE-2022-50062

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: bgmac: Fix a BUG triggered by wrong bytes_compl<br /> <br /> On one of our machines we got:<br /> <br /> kernel BUG at lib/dynamic_queue_limits.c:27!<br /> Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM<br /> CPU: 0 PID: 1166 Comm: irq/41-bgmac Tainted: G W O 4.14.275-rt132 #1<br /> Hardware name: BRCM XGS iProc<br /> task: ee3415c0 task.stack: ee32a000<br /> PC is at dql_completed+0x168/0x178<br /> LR is at bgmac_poll+0x18c/0x6d8<br /> pc : [] lr : [] psr: 800a0313<br /> sp : ee32be14 ip : 000005ea fp : 00000bd4<br /> r10: ee558500 r9 : c0116298 r8 : 00000002<br /> r7 : 00000000 r6 : ef128810 r5 : 01993267 r4 : 01993851<br /> r3 : ee558000 r2 : 000070e1 r1 : 00000bd4 r0 : ee52c180<br /> Flags: Nzcv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none<br /> Control: 12c5387d Table: 8e88c04a DAC: 00000051<br /> Process irq/41-bgmac (pid: 1166, stack limit = 0xee32a210)<br /> Stack: (0xee32be14 to 0xee32c000)<br /> be00: ee558520 ee52c100 ef128810<br /> be20: 00000000 00000002 c0116298 c04b5a18 00000000 c0a0c8c4 c0951780 00000040<br /> be40: c0701780 ee558500 ee55d520 ef05b340 ef6f9780 ee558520 00000001 00000040<br /> be60: ffffe000 c0a56878 ef6fa040 c0952040 0000012c c0528744 ef6f97b0 fffcfb6a<br /> be80: c0a04104 2eda8000 c0a0c4ec c0a0d368 ee32bf44 c0153534 ee32be98 ee32be98<br /> bea0: ee32bea0 ee32bea0 ee32bea8 ee32bea8 00000000 c01462e4 ffffe000 ef6f22a8<br /> bec0: ffffe000 00000008 ee32bee4 c0147430 ffffe000 c094a2a8 00000003 ffffe000<br /> bee0: c0a54528 00208040 0000000c c0a0c8c4 c0a65980 c0124d3c 00000008 ee558520<br /> bf00: c094a23c c0a02080 00000000 c07a9910 ef136970 ef136970 ee30a440 ef136900<br /> bf20: ee30a440 00000001 ef136900 ee30a440 c016d990 00000000 c0108db0 c012500c<br /> bf40: ef136900 c016da14 ee30a464 ffffe000 00000001 c016dd14 00000000 c016db28<br /> bf60: ffffe000 ee21a080 ee30a400 00000000 ee32a000 ee30a440 c016dbfc ee25fd70<br /> bf80: ee21a09c c013edcc ee32a000 ee30a400 c013ec7c 00000000 00000000 00000000<br /> bfa0: 00000000 00000000 00000000 c0108470 00000000 00000000 00000000 00000000<br /> bfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000<br /> bfe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000<br /> [] (dql_completed) from [] (bgmac_poll+0x18c/0x6d8)<br /> [] (bgmac_poll) from [] (net_rx_action+0x1c4/0x494)<br /> [] (net_rx_action) from [] (do_current_softirqs+0x1ec/0x43c)<br /> [] (do_current_softirqs) from [] (__local_bh_enable+0x80/0x98)<br /> [] (__local_bh_enable) from [] (irq_forced_thread_fn+0x84/0x98)<br /> [] (irq_forced_thread_fn) from [] (irq_thread+0x118/0x1c0)<br /> [] (irq_thread) from [] (kthread+0x150/0x158)<br /> [] (kthread) from [] (ret_from_fork+0x14/0x24)<br /> Code: a83f15e0 0200001a 0630a0e1 c3ffffea (f201f0e7)<br /> <br /> The issue seems similar to commit 90b3b339364c ("net: hisilicon: Fix a BUG<br /> trigered by wrong bytes_compl") and potentially introduced by commit<br /> b38c83dd0866 ("bgmac: simplify tx ring index handling").<br /> <br /> If there is an RX interrupt between setting ring-&gt;end<br /> and netdev_sent_queue() we can hit the BUG_ON as bgmac_dma_tx_free()<br /> can miscalculate the queue size while called from bgmac_poll().<br /> <br /> The machine which triggered the BUG runs a v4.14 RT kernel - but the issue<br /> seems present in mainline too.