Date: Mon, 29 Dec 2025 12:51:03 -0800 From: Gleb Smirnoff <glebius@freebsd.org> To: Alastair Hogge <agh@riseup.net> Cc: A FreeBSD User <freebsd@walstatt-de.de>, Ronald Klop <ronald-lists@klop.ws>, FreeBSD CURRENT <freebsd-current@freebsd.org>, David Wolfskill <david@catwhisker.org> Subject: Re: CURRENT: kernel panic in IPFW while stopping jails Message-ID: <aVLpt7RTFQbJFVxJ@cell.glebi.us> In-Reply-To: <291e26bfbe2b51835f7672db3a2e3593@riseup.net> References: <20251225170828.7aef61df@hermann> <902742484.3865.1766683845222@localhost> <20251225190836.6769e6d6@hermann> <20251226103308.72204662@thor.sb211.local> <291e26bfbe2b51835f7672db3a2e3593@riseup.net>
index | next in thread | previous in thread | raw e-mail
Alastair, thanks for running bisect. I see that you started it from a revision that already has 129e15d4994311958db59a1718d4ff42d440ce2b and it still is bad for you. This means a second problem exists in 3daae1ac1d82ecdcd855101bab5206e914b12350. I just tried to reproduce with exactly Dan's ruleset - no panic :( Maybe you can share how you reproduced during your bisection? Or maybe you can dump a core and share it? Very sorry for this breakage during the holidays. I really want to fix that ASAP, but I need more data. On Sun, Dec 28, 2025 at 11:36:05AM +0000, Alastair Hogge wrote: A> $ git bisect log A> git bisect start A> # status: waiting for both good and bad commits A> # bad: [086bedb11a853801e82234b8a1a64f0df52d9e52] tools.build: also add A> sys/_visible.h to SYSINCS A> git bisect bad 086bedb11a853801e82234b8a1a64f0df52d9e52 A> # status: waiting for good commit(s), bad commit known A> # good: [44cb1e857f048d2326bdc1a032ccd2c04d2bcdc9] tcp: improve A> credential handling in syncache A> git bisect good 44cb1e857f048d2326bdc1a032ccd2c04d2bcdc9 A> # good: [b0c7eaf83d21bbc333e247ab9e136965b3ca54ed] bhyve/slirp: Drop A> privileges before entering capability mode A> git bisect good b0c7eaf83d21bbc333e247ab9e136965b3ca54ed A> # good: [6a75e3951506c12b42428a47710d07cadcdd723e] ofed/libibverbs: A> remove strdupa() hack from config.h A> git bisect good 6a75e3951506c12b42428a47710d07cadcdd723e A> # bad: [1fad49baf390cb52f238e6c352d0bc0893c008c3] sdhci: Try to complete A> the last transaction if dumping A> git bisect bad 1fad49baf390cb52f238e6c352d0bc0893c008c3 A> # good: [9d9974457ce8c6cf9023884ab457d4712dcc237f] bhyvectl: fix build A> without BHYVE_SNAPSHOT A> git bisect good 9d9974457ce8c6cf9023884ab457d4712dcc237f A> # bad: [52395203f9ac40d321ed55d93e9887300261d3bf] MFV: Import blocklist A> 2025-12-15 (8a4b011) A> git bisect bad 52395203f9ac40d321ed55d93e9887300261d3bf A> # good: [c112ad75605ccdfcb8bbce2f57b0e7a077f057f8] options: describe A> WITH_IPFILTER_IPFS A> git bisect good c112ad75605ccdfcb8bbce2f57b0e7a077f057f8 A> # good: [8774a990ee4094f16d596d4b78e0f3239e5d0c88] bpf: modularize A> ifnet(9) part of bpf A> git bisect good 8774a990ee4094f16d596d4b78e0f3239e5d0c88 A> # bad: [1615eff94cda8619561b73186ec8098cc8b14c5c] usb: don't create A> ifnet(9) for usbus devices A> git bisect bad 1615eff94cda8619561b73186ec8098cc8b14c5c A> # good: [ddf4f9eda9c295082f17e7f26963666b72c97bb9] ipfw: create "ipfw0" A> and "ipfwlog0" bpf tapping points without ifnet(9) A> git bisect good ddf4f9eda9c295082f17e7f26963666b72c97bb9 A> # bad: [3daae1ac1d82ecdcd855101bab5206e914b12350] ipfw: create a bpf tap A> point for every log rule A> git bisect bad 3daae1ac1d82ecdcd855101bab5206e914b12350 A> # good: [1c5021f5251b231b614ad9cd175bcb4250495c12] ifconfig: print A> warning and return success on ipfw0, ipfwlog0 cloning A> git bisect good 1c5021f5251b231b614ad9cd175bcb4250495c12 A> # first bad commit: [3daae1ac1d82ecdcd855101bab5206e914b12350] ipfw: A> create a bpf tap point for every log rule A> A> https://codeberg.org/FreeBSD/freebsd-src/commit/3daae1ac1d82ecdcd855101bab5206e914b12350 A> ipfw: create a bpf tap point for every log rule A> A> Dynamically allocate bpf tap points for every rule that has "log". A> The name is "ipfw%u", where %u is substituted to the rule number. A> The default catch all "ipfw0" tap still exists for compatibility A> and it will catch packets in case if there are no bpf listeners A> on a per-rule tap. A> A> Reviewed by: ae A> Differential Revision: https://reviews.freebsd.org/D53877 -- Gleb Smirnoffhome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?aVLpt7RTFQbJFVxJ>