Rob Norris
Rob Norris
No change, sigh. At least that rules out misuse of `bio_put()`. `zpool_reopen/zpool_reopen_003_pos` is a reliable trigger, so my next act is to go through it step by step and try...
I've been working on this sporadically for the last month. Definitely made progress, though not solved yet. The `bio_chain()` mechanism is very good, and simplifies our code - no need...
Alright, I'm kinda baffled at this point. I've narrowed things down a long way but I can't come up with a workable theory. The test in question is `cli_root/zpool_reopen/zpool_reopen_003_pos.ksh`. It...
Little breakthrough today. If I comment out the "move back compound head" section, everything is solid. Turns out before ~4.5, head and tail pages are refcounted separately (see torvalds/linux@ddc58f2). So...
If this turns out to be true, then my rough guess at what's happening is that the head page is entered into the bio, which takes a reference. After the...
Sigh this thing. What compiler version?
> @gititdone876 Should I include the logs in this comment, or edit my original post? Probably doesn't matter too much; personally I'd add to the original post with a line...
@gititdone876 thank you. "oopsie" :) So, I'm not totally sure what's happening, because this particular kernel version (6.7) is responding in a way I can quite see my way through....
Retested with 6.9-rc7 (likely final 6.9-rc), no change.
I did look into this a bit last night, though ultimately didn't get anywhere. There's more change in Linux in `block/bdev.c` in Linux in 6.8->6.9 than I realised; it's not...