Hi Nick, Nick Black wrote: > I believe this bug has been fixed, but am not certain. I was > able to reproduce a crash pretty trivially, but it didn't have > anything to do with pressing F1 (which I was able to do just > fine, if growlight started correctly).
Hmmmm. The readline rltty.c patch you sent me (and mentioned at the end of your mail, sounded as if you found an issue elsewhere? But it is unrelated? > The root cause of the segfault I was running into was a /proc/mounts > larger than a page, and a broken function for reading such files. > The broken function left us with an empty mount map, which led to > problems. I'm quite sure I had not that issue. Mostly because I had tons of block devices shown by growlight. Especially now that I have seen https://github.com/dankamongmen/growlight/issues/103, I don't think I had any of the output there, sorry. (Still on Kernel 5.7.10, btw.) Nevertheless my /proc/mounts is not that small: $ wc /proc/mounts 49 294 3730 /proc/mounts But IIRC, a page is like 4k or so. So it's still below it, right? > I've mapped F1 to 'H' so that it now toggles the Help screen. Ok, so I tried Shift-H: And indeed, there is a help box. Yay! (Would be great to put at least _that_ key binding in the man page, maybe all shown by it. Would have helped me a lot and would also avoid data loss by trying out keybindings as I see that some of them are really dangerous to just try out. I hope I didn't do any damage to my system.) I'm not sure what all I did before and after displaying the help, but I got it segfaulting again, but I got it segfaulting on exit by pressing "q": 10814 renders, 5.29s total (138.86µs min, 21.41ms max, 489.17µs avg) 1.57MiB total (0.00B min, 589.12KiB max, 0.149KiB avg) 2044.3 theoretical FPS, 0 failed renders RGB emits:elides: def 85490:60740 fg 87196:1773 bg 61:1782 Elide rates: 41.54% 1.99% 96.69% Cell emits:elides: 148073/88710387 (99.83%) free(): double free detected in tcache 2 [1] 22757 abort (core dumped) growlight growlight 21.05s user 0.10s system 8% cpu 3:57.44 total I did at least the following, but nothing of that again caused it to segfault, so I currently can't reproduce that crash: * making the uxterm fullsize using my window manager (i3) * then displaying the help window by pressing Shift-H * then undoing fullsize again (the help window vanished) * pressing H again (nothing happened) * pressing H yet again (help window is there again) * making the uxterm fullsize again (help window is about at middle height) * undoing fullsize again * maybe some more fullsizing again and undoing, don't remember. * Pressing "q" for exiting. > If you have the time and inclination, I'd be indebted were you > to test from growlight head, and see if your problem is still > there. Did that. > If you can't, that's perfectly understandable, Well, yes and no. :-) Yes, I can't reproduce the originally reported issue by pressing F1. F1 now shows the help. Yes! Thanks! Buuuuuut: The originally reported issue (crash by segfault) is still there if I press another (unassigned) F-key like e.g. F2. Tried all F-keys from F2 to F12 and it segfaults on all of those — after checking the help that this won't damage my system. :-) So this is probably a more generic issue. > While I was there, I also fixed an issue that was causing > growlight-readline to segfault out under certain scenarios, which > led to a bug report + patch for libreadline (never thought I'd > manage one of those). Hehe. > Thank you for this report, Axel, and your patience! You're welcome! Regards, Axel -- ,''`. | Axel Beckert <a...@debian.org>, https://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `- | 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE