Just adding a few datapoints from crashstats: - these are all content process crashes (of course, because of `~ContenChild`). - The first build ID I see this starting with seems to be 20240723211328 for Firefox and 20240724215903 for Fenix (both nightly). - Bug 1834864 landed just a few days earlier but was immediately uplifted to beta 129, so that seems to not fit with beta crashes (and I think that was the bug I had vaguely in mind pinging :nika for, though she might spot something else, of course) - The first Fenix beta crashes seem to arrive with beta 5. [I cannot spot anything outstanding in this push range](https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_130_0b4_RELEASE&tochange=FIREFOX_130_0b5_RELEASE), though. And also the earlier 130 beta uplifts seem unrelated to me at a first glance.
Bug 1915788 Comment 3 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
Just adding a few more datapoints: - these are all content process crashes (of course, because of `~ContenChild`). - The first build ID I see this starting with seems to be 20240723211328 for Firefox and 20240724215903 for Fenix (both nightly). - Bug 1834864 landed just a few days earlier but was immediately uplifted to beta 129, so that seems to not fit with beta crashes - Bug 1728331 looks pretty relevant for what it does but has been fixed already in 129, too. - The first Fenix beta crashes seem to arrive with beta 5. [I cannot spot anything outstanding in this push range](https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_130_0b4_RELEASE&tochange=FIREFOX_130_0b5_RELEASE), though. And also the earlier 130 beta uplifts seem unrelated to me at a first glance. - on the incident channel, :aryx said that these are all startup crashes (wrt to trhe single process). The stack trace indicates that we see [`XRE_InitChildProcess`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#651) arriving at its very end, which normally should never happen as IIRC we run our shutdown and exit(0) on top of the message loop. The only case where this can happen seems to be [a failure during `process->Init(...)` ](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#612-614) which would immediately return and thus destroy [`UniquePtr<ProcessChild> process;`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#548). So it appears to me that we encounter an error during process initialization and that will make the dtor angry (for probably no good reason in that case). Bug 1471720 landed July 23 and [introduced a new failure case](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/xpcom/build/XPCOMInit.cpp#391-394) during startup, if the omnijar is corrupted. I would expect to see for Fenix the corresponding parent process crashes on bug 1909700, but I do not see them. I would assume that these crashes are just a symptom of a corrupted installation.
Just adding a few more datapoints: - these are all content process crashes (of course, because of `~ContenChild`). - The first build ID I see this starting with seems to be 20240723211328 for Firefox and 20240724215903 for Fenix (both nightly). - Bug 1834864 landed just a few days earlier but was immediately uplifted to beta 129, so that seems to not fit with beta crashes - Bug 1728331 ~~looks pretty relevant for what it does~~ but has been fixed already in 129, too. - The first Fenix beta crashes seem to arrive with beta 5. [I cannot spot anything outstanding in this push range](https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_130_0b4_RELEASE&tochange=FIREFOX_130_0b5_RELEASE), though. And also the earlier 130 beta uplifts seem unrelated to me at a first glance. - on the incident channel, :aryx said that these are all startup crashes (wrt to trhe single process). The stack trace indicates that we see [`XRE_InitChildProcess`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#651) arriving at its very end, which normally should never happen as IIRC we run our shutdown and exit(0) on top of the message loop. The only case where this can happen seems to be [a failure during `process->Init(...)` ](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#612-614) which would immediately return and thus destroy [`UniquePtr<ProcessChild> process;`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#548). So it appears to me that we encounter an error during process initialization and that will make the dtor angry (for probably no good reason in that case). Bug 1471720 landed July 23 and [introduced a new failure case](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/xpcom/build/XPCOMInit.cpp#391-394) during startup, if the omnijar is corrupted. I would expect to see for Fenix the corresponding parent process crashes on bug 1909700, but I do not see them. I would assume that these crashes are just a symptom of a corrupted installation.
Just adding a few more datapoints: - these are all content process crashes (of course, because of `~ContenChild`). - The first build ID I see this starting with seems to be 20240723211328 for Firefox and 20240724215903 for Fenix (both nightly). - Bug 1834864 landed just a few days earlier but was immediately uplifted to beta 129, so that seems to not fit with beta crashes - Bug 1728331 ~~looks pretty relevant for what it does~~ but has been fixed already in 129, too. - The first Fenix beta crashes seem to arrive with beta 5. [I cannot spot anything outstanding in this push range](https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_130_0b4_RELEASE&tochange=FIREFOX_130_0b5_RELEASE), though. And also the earlier 130 beta uplifts seem unrelated to me at a first glance. - on the incident channel, :aryx said that these are all startup crashes (wrt to trhe single process). The stack trace indicates that we see [`XRE_InitChildProcess`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#651) arriving at its very end, which normally should never happen as IIRC we run our shutdown and exit(0) on top of the message loop. The only case where this can happen seems to be [a failure during `process->Init(...)` ](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#612-614) which would immediately return and thus destroy [`UniquePtr<ProcessChild> process;`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#548). So it appears to me that we encounter an error during process initialization and that will make the dtor angry (for probably no good reason in that case). Bug 1471720 landed July 23 and [introduced a new failure case](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/xpcom/build/XPCOMInit.cpp#391-394) during startup if the omnijar is corrupted. I would expect to see for Fenix the corresponding parent process crashes on bug 1909700, but I do not see them. I would assume that these crashes are just a symptom of a corrupted installation.
Just adding a few more datapoints: - these are all content process crashes (of course, because of `~ContenChild`). - The first build ID I see this starting with seems to be 20240723211328 for Firefox and 20240724215903 for Fenix (both nightly). - Bug 1834864 landed just a few days earlier but was immediately uplifted to beta 129, so that seems to not fit with beta crashes - Bug 1728331 ~~looks pretty relevant for what it does~~ but has been fixed already in 129, too. - The first Fenix beta crashes seem to arrive with beta 5. [I cannot spot anything outstanding in this push range](https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=FIREFOX_130_0b4_RELEASE&tochange=FIREFOX_130_0b5_RELEASE), though. And also the earlier 130 beta uplifts seem unrelated to me at a first glance. - on the incident channel, :aryx said that these are all startup crashes (wrt to the single process). The stack trace indicates that we see [`XRE_InitChildProcess`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#651) arriving at its very end, which normally should never happen as IIRC we run our shutdown and exit(0) on top of the message loop. The only case where this can happen seems to be [a failure during `process->Init(...)` ](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#612-614) which would immediately return and thus destroy [`UniquePtr<ProcessChild> process;`](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/toolkit/xre/nsEmbedFunctions.cpp#548). So it appears to me that we encounter an error during process initialization and that will make the dtor angry (for probably no good reason in that case). Bug 1471720 landed July 23 and [introduced a new failure case](https://searchfox.org/mozilla-central/rev/cb5faf5dd5176494302068c553da97b4d08aa339/xpcom/build/XPCOMInit.cpp#391-394) during startup if the omnijar is corrupted. I would expect to see for Fenix the corresponding parent process crashes on bug 1909700, but I do not see them. I would assume that these crashes are just a symptom of a corrupted installation.