bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
On Sun, Apr 5, 2020 at 9:29 AM Ludovic Courtès wrote: > Hi, > > Marnen Laibow-Koser skribis: > > > There’s now an lldb backtrace at > > https://gitlab.com/marnen/lilypond-mac-builder/issues/16#note_317315811 > . > > Hopefully it’s somewhat useful? > > Sorry, I can’t make much sense of it. :-/ > What would be more helpful? > > Ludo’. > Best, -- Marnen Laibow-Koser mar...@marnen.org http://www.marnen.org
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
Hi, Marnen Laibow-Koser skribis: > There’s now an lldb backtrace at > https://gitlab.com/marnen/lilypond-mac-builder/issues/16#note_317315811 . > Hopefully it’s somewhat useful? Sorry, I can’t make much sense of it. :-/ Ludo’.
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
On Sat, Mar 7, 2020 at 3:36 PM Ludovic Courtès wrote: > Hi, > > Marnen Laibow-Koser skribis: > > > On Sat, Mar 7, 2020 at 10:12 AM Ludovic Courtès wrote: > > [...] > > >> Could you provide a backtrace of the segfault? > >> > > > > I’m not sure how I would be able to get any more detailed debugging info > > than the logs that I posted in the GitLab issue. Have you looked at > > those? Alternatively, is there an environment variable or anything that > > would get Guile to provide more information when it does segfault? > > I would just run GDB on the dumped core, and run “bt” there (apologies > if that info is already in the GitLab issue, I didn’t see it.) There’s now an lldb backtrace at https://gitlab.com/marnen/lilypond-mac-builder/issues/16#note_317315811 . Hopefully it’s somewhat useful? Note that this is from a second affected user, who said that the issue went away for him after he reinstalled macOS (Catalina, FWIW). Best, -- Marnen Laibow-Koser mar...@marnen.org http://www.marnen.org Sent from Gmail Mobile
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
On Sat, Mar 7, 2020 at 3:36 PM Ludovic Courtès wrote: > Hi, > > Marnen Laibow-Koser skribis: > > > On Sat, Mar 7, 2020 at 10:12 AM Ludovic Courtès wrote: > > [...] > > >> Could you provide a backtrace of the segfault? > >> > > > > I’m not sure how I would be able to get any more detailed debugging info > > than the logs that I posted in the GitLab issue. Have you looked at > > those? Alternatively, is there an environment variable or anything that > > would get Guile to provide more information when it does segfault? > > I would just run GDB on the dumped core, and run “bt” there (apologies > if that info is already in the GitLab issue, I didn’t see it.) It isn’t. I rarely have to do this sort of debugging, so I didn’t really know how to do it. Let me see what I can do, though I don’t know if there was an actual coredump. > > >> Is it a recent regression of your LilyPond package for macOS? > > > > > > I’m afraid that question doesn’t have much meaning either: this packaging > > is newly developed, so there’s nothing to regress *from*. > > Ah OK. Then perhaps you could look at how Homebrew, MacPorts & > co. package LilyPond in search of inspiration (speaking with my distro > hacker hat on :-))? I did (you can see the lilypond-dev list archives if you’re really interested in the gory details). This packaging is the result of the process of lots of research on how to actually make it all work for 64-bit Mac OS—and it does work, except, consistently, for one particular user. > > Thanks, > Ludo’. Best, -- Marnen Laibow-Koser mar...@marnen.org http://www.marnen.org Sent from Gmail Mobile
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
Hi, Marnen Laibow-Koser skribis: > On Sat, Mar 7, 2020 at 10:12 AM Ludovic Courtès wrote: [...] >> Could you provide a backtrace of the segfault? >> > > I’m not sure how I would be able to get any more detailed debugging info > than the logs that I posted in the GitLab issue. Have you looked at > those? Alternatively, is there an environment variable or anything that > would get Guile to provide more information when it does segfault? I would just run GDB on the dumped core, and run “bt” there (apologies if that info is already in the GitLab issue, I didn’t see it.) >> Is it a recent regression of your LilyPond package for macOS? > > > I’m afraid that question doesn’t have much meaning either: this packaging > is newly developed, so there’s nothing to regress *from*. Ah OK. Then perhaps you could look at how Homebrew, MacPorts & co. package LilyPond in search of inspiration (speaking with my distro hacker hat on :-))? Thanks, Ludo’.
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
On Sat, Mar 7, 2020 at 10:12 AM Ludovic Courtès wrote: > Hi, > > Marnen Laibow-Koser skribis: > > > I'm not sure this is a bug report yet, but it may become one. I'm having > > what looks like a Guile segfault issue for some Mac OS Mojave (10.14) > > users. I've been packaging 64-bit Mac builds of LilyPond, which includes > > libguile 1.8. The builds work well for myself and for nearly everyone > else, > > but one user (on Mojave, as am I) consistently reports a segfault right > > around the time libguile is loaded; see > > https://gitlab.com/marnen/lilypond-mac-builder/issues/16 . The LilyPond > > group hasn't provided much help on figuring out this particular issue, > so I > > thought I'd come over here... > > > > There's more information in the GitLab issue I mentioned above, but > please > > let me know what else would be helpful in diagnosing and fixing this > > particular problem. (Upgrading to a recent Guile is not an option right > > now AFAIK: LilyPond barely works with Guile 2, let alone 3, and fixing > > *that* would probably be a much larger issue.) > > Could you provide a backtrace of the segfault? > I’m not sure how I would be able to get any more detailed debugging info than the logs that I posted in the GitLab issue. Have you looked at those? Alternatively, is there an environment variable or anything that would get Guile to provide more information when it does segfault? > Does the problem occur on GNU/Linux as well? > I have no way of knowing. I discovered this issue in the course of figuring out how to package LilyPond into a 64-bit Mac .app bundle. I’ve never run this particular build process on Linux, and I’m not sure that it would even be meaningful to do so given that the primary point of it is to make a Mac .app. > Is it a recent regression of your LilyPond package for macOS? I’m afraid that question doesn’t have much meaning either: this packaging is newly developed, so there’s nothing to regress *from*. What I can tell you is that it consistently affects one user (and, I believe, on all builds I’ve so far done with this method), but everyone else has no problems with these builds, and the user who *is* having problems was able to run the 32-bit Mac builds, but those are built by a somewhat different process. I know this information is less helpful and comprehensive than we might like, but since I’m still developing this particular build and packaging process, there’s very little to go on. > > Unfortunately, Guile 1.8 has been unmaintained for several years and its > code base has almost nothing in common with that of 2.0 and later, so I > won’t make any promises. Understood. Believe me, if I could upgrade to a recent Guile, I would, but my objective here is mostly to get a working 64-bit Mac build of LilyPond while changing the code as little as possible. :) > > Thanks, > Ludo’. > -- Marnen Laibow-Koser mar...@marnen.org http://www.marnen.org Sent from Gmail Mobile
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
Hi, Marnen Laibow-Koser skribis: > I'm not sure this is a bug report yet, but it may become one. I'm having > what looks like a Guile segfault issue for some Mac OS Mojave (10.14) > users. I've been packaging 64-bit Mac builds of LilyPond, which includes > libguile 1.8. The builds work well for myself and for nearly everyone else, > but one user (on Mojave, as am I) consistently reports a segfault right > around the time libguile is loaded; see > https://gitlab.com/marnen/lilypond-mac-builder/issues/16 . The LilyPond > group hasn't provided much help on figuring out this particular issue, so I > thought I'd come over here... > > There's more information in the GitLab issue I mentioned above, but please > let me know what else would be helpful in diagnosing and fixing this > particular problem. (Upgrading to a recent Guile is not an option right > now AFAIK: LilyPond barely works with Guile 2, let alone 3, and fixing > *that* would probably be a much larger issue.) Could you provide a backtrace of the segfault? Does the problem occur on GNU/Linux as well? Is it a recent regression of your LilyPond package for macOS? Unfortunately, Guile 1.8 has been unmaintained for several years and its code base has almost nothing in common with that of 2.0 and later, so I won’t make any promises. Thanks, Ludo’.
bug#39863: Guile 1.8, apparent segfault on some Macs running Mojave
Hi there! I'm not sure this is a bug report yet, but it may become one. I'm having what looks like a Guile segfault issue for some Mac OS Mojave (10.14) users. I've been packaging 64-bit Mac builds of LilyPond, which includes libguile 1.8. The builds work well for myself and for nearly everyone else, but one user (on Mojave, as am I) consistently reports a segfault right around the time libguile is loaded; see https://gitlab.com/marnen/lilypond-mac-builder/issues/16 . The LilyPond group hasn't provided much help on figuring out this particular issue, so I thought I'd come over here... There's more information in the GitLab issue I mentioned above, but please let me know what else would be helpful in diagnosing and fixing this particular problem. (Upgrading to a recent Guile is not an option right now AFAIK: LilyPond barely works with Guile 2, let alone 3, and fixing *that* would probably be a much larger issue.) Best, -- Marnen Laibow-Koser mar...@marnen.org http://www.marnen.org