I run CURRENT revision baef3a5b585f. my kernel config is: include GENERIC ident VAL nodevice em nodevice ixl nodevice iavf nodevice ice nodevice ix nodevice ixv When I try to kldunload if_ice, I get: panic: _sx_xlock_hard: recursed on non-recursive sx iflib ctx lock @ /usr/src/sys/net/iflib.c:2582 cpuid = 3 time = 1685630930 KDB: stack backtrace: db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 0xfffffe010b8475d0 vpanic() at vpanic+0x150/frame 0xfffffe010b847620 panic() at panic+0x43/frame 0xfffffe010b847680 _sx_xlock_hard() at _sx_xlock_hard+0x82d/frame 0xfffffe010b847720 _sx_xlock() at _sx_xlock+0xb3/frame 0xfffffe010b847760 iflib_media_status() at iflib_media_status+0x32/frame 0xfffffe010b847790 ifmedia_ioctl() at ifmedia_ioctl+0x16a/frame 0xfffffe010b8477c0 dump_iface() at dump_iface+0x12f/frame 0xfffffe010b847840 rtnl_handle_ifevent() at rtnl_handle_ifevent+0xab/frame 0xfffffe010b8478c0 if_attach_internal() at if_attach_internal+0x41e/frame 0xfffffe010b847910 ether_ifattach() at ether_ifattach+0x29/frame 0xfffffe010b847940 iflib_device_register() at iflib_device_register+0xbf7/frame 0xfffffe010b8479d0 iflib_device_attach() at iflib_device_attach+0xb5/frame 0xfffffe010b847a00 device_attach() at device_attach+0x3c0/frame 0xfffffe010b847a40 device_probe_and_attach() at device_probe_and_attach+0x70/frame 0xfffffe010b847a70 pci_driver_added() at pci_driver_added+0xf2/frame 0xfffffe010b847ab0 devclass_driver_added() at devclass_driver_added+0x39/frame 0xfffffe010b847af0 device_do_deferred_actions() at device_do_deferred_actions+0x3a/frame 0xfffffe010b847b10 devctl2_ioctl() at devctl2_ioctl+0x531/frame 0xfffffe010b847be0 devfs_ioctl() at devfs_ioctl+0xcc/frame 0xfffffe010b847c30 vn_ioctl() at vn_ioctl+0xc0/frame 0xfffffe010b847ca0 devfs_ioctl_f() at devfs_ioctl_f+0x1e/frame 0xfffffe010b847cc0 kern_ioctl() at kern_ioctl+0x1fe/frame 0xfffffe010b847d30 sys_ioctl() at sys_ioctl+0x154/frame 0xfffffe010b847e00 amd64_syscall() at amd64_syscall+0x140/frame 0xfffffe010b847f30 fast_syscall_common() at fast_syscall_common+0xf8/frame 0xfffffe010b847f30 --- syscall (54, FreeBSD ELF64, ioctl), rip = 0x1da51d7cee6a, rsp = 0x1da51b4eb188, rbp = 0x1da51b4eb240 ---
This issue also happens if I disable devmatch and load the ice module manually. For some reason it doesn't happen during autoloading with enabled devmatch.
I added "devmatch_blocklist="if_ice"" in /etc/rc.conf and automatically reload doesn't occur. It helps to confirm that panic exists only during load driver. It looks like the CTX_LOCK is already locked in iflib_media_status. I remember that a few weeks ago disabling NETLINK was helping but not now.
CTX is locked first in iflib_device_register and second in iflib_media_status(here panic). SIOCGIFMEDIA is received in iflib_if_ioctl(then ifmedia_ioctl -> iflib_media_status) but on 13.2 it is not sent from get_operstate_ether() in netlink module. On 14.0 is sent from netlink. Can issue be related to ifnet_arrival_event? It needs to be investigated more.
Prepared fix for this: https://reviews.freebsd.org/D40557
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/src/commit/?id=a52f23f4c49e4766fb9eb0bf460cc83c5f63f17d commit a52f23f4c49e4766fb9eb0bf460cc83c5f63f17d Author: Eric Joyner <erj@FreeBSD.org> AuthorDate: 2023-07-19 22:40:46 +0000 Commit: Eric Joyner <erj@FreeBSD.org> CommitDate: 2023-07-19 22:51:26 +0000 iflib: Unlock ctx lock around call to ether_ifattach() Panic occurs during loading driver using kldload. It exists since netlink is enabled. There is problem with double locking ctx. This fix allows to call ether_ifattach() without locked ctx. Signed-off-by: Eric Joyner <erj@FreeBSD.org> PR: 271768 Reviewed by: erj@, jhb@ MFC after: 1 day Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D40557 sys/net/iflib.c | 6 ++++++ 1 file changed, 6 insertions(+)
A commit in branch stable/13 references this bug: URL: https://cgit.FreeBSD.org/src/commit/?id=93f39b4675865094b298e1fca5bcfa788348dd36 commit 93f39b4675865094b298e1fca5bcfa788348dd36 Author: Przemyslaw Lewandowski <przemyslawx.lewandowski@intel.com> AuthorDate: 2023-07-19 22:40:46 +0000 Commit: Eric Joyner <erj@FreeBSD.org> CommitDate: 2023-07-20 22:42:22 +0000 iflib: Unlock ctx lock around call to ether_ifattach() Panic occurs during loading driver using kldload. It exists since netlink is enabled. There is problem with double locking ctx. This fix allows to call ether_ifattach() without locked ctx. Signed-off-by: Eric Joyner <erj@FreeBSD.org> PR: 271768 Reviewed by: erj@, jhb@ Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D40557 (cherry picked from commit a52f23f4c49e4766fb9eb0bf460cc83c5f63f17d) sys/net/iflib.c | 6 ++++++ 1 file changed, 6 insertions(+)
A commit in branch stable/12 references this bug: URL: https://cgit.FreeBSD.org/src/commit/?id=2617bca5a8c98e9e84fba3f2ebd9302e8e893b8f commit 2617bca5a8c98e9e84fba3f2ebd9302e8e893b8f Author: Przemyslaw Lewandowski <przemyslawx.lewandowski@intel.com> AuthorDate: 2023-07-19 22:40:46 +0000 Commit: Eric Joyner <erj@FreeBSD.org> CommitDate: 2023-07-20 22:56:41 +0000 iflib: Unlock ctx lock around call to ether_ifattach() Panic occurs during loading driver using kldload. It exists since netlink is enabled. There is problem with double locking ctx. This fix allows to call ether_ifattach() without locked ctx. Signed-off-by: Eric Joyner <erj@FreeBSD.org> PR: 271768 Reviewed by: erj@, jhb@ Sponsored by: Intel Corporation Differential Revision: https://reviews.freebsd.org/D40557 (cherry picked from commit a52f23f4c49e4766fb9eb0bf460cc83c5f63f17d) sys/net/iflib.c | 6 ++++++ 1 file changed, 6 insertions(+)
At least at Intel, Przemyslaw's fix did fix this issue for us.