[Bug 280037] KTLS with Intel QAT may trigger kernel panics
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Wed, 01 Jan 2025 23:51:37 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=280037 --- Comment #16 from ss3bsd <3226388001@jcom.home.ne.jp> --- Created attachment 256331 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=256331&action=edit draft patch that addresses comment no.11 and 14 I have tried to create a draft patch (attached) for comment no.11 (resource leak) and 14 (panic). With high TX load test & kldunload qat, Before applying this patch (for no.11): qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: waiting for session un-busy failed qat_ocf0: unable to remove encrypt session qat_ocf0: detached Warning: memory type qat_ocf leaked memory on destroy (250 allocations, 1024000 bytes leaked). qat0: qat_dev0 stopped 6 acceleration engines qat0: Resetting device qat_dev0 qat0: detached After (EAGAIN logs are added this time for debug): % dmesg | tail qat_ocf0: EAGAIN cpaCySymDpEnqueueOp qat_ocf0: EAGAIN cpaCySymDpEnqueueOp ktls_ocf_dispatch_async_cb EAGAIN DELAY ktls_ocf_dispatch_async_cb EAGAIN DELAY qat_ocf0: EAGAIN cpaCySymDpEnqueueOp ktls_ocf_dispatch_async_cb EAGAIN DELAY qat_ocf0: EAGAIN cpaCySymDpEnqueueOp qat_ocf0: EAGAIN cpaCySymDpEnqueueOp ktls_ocf_dispatch_async_cb EAGAIN DELAY ktls_ocf_dispatch_async_cb EAGAIN DELAY qat_ocf0: detached qat0: qat_dev0 stopped 6 acceleration engines qat0: Resetting device qat_dev0 qat0: detached -- You are receiving this mail because: You are the assignee for the bug.