On Sat, Apr 05, 2025 at 04:45:50PM -0700, Luck, Tony wrote: > Several email threads on this over the past few days, but none > seem to be getting much attention: > > https://lore.kernel.org/all/?q=%22%28%26q-%3Eelevator_lock%29%7B%2B.%2B.%7D-%7B4%3A4%7D%2C+at%3A+elv_iosched_store%22 > > Since tomorrow is "-rc1" day I thought I'd add my lockdep splat to the > mix. > > System is a 2-socket Icelake. Kernel built from: > > f4d2ef48250a ("Merge tag 'kbuild-v6.15' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild") > > Though I saw the same thing a few days ago, so the problem was > introduced much earlier in the merge window. Bisecting leads to this: ffa1e7ada456 ("block: Make request_queue lockdep splats show up earlier") Which may be doing what it says on the tin. But, I don't think I've ever seen the "type 1 splats" on this machine even though every kernel I build has lockdep enabled. So it this showing up a real problem that happens to be very difficult to hit? Or has it introduced a false positive? -Tony
