Re: Bus error due to v9 VIS on v8 ?
vom513 <[email protected]>
| Newsgroups | gmane.os.netbsd.ports.sparc |
|---|---|
| Message-ID | <[email protected]> |
> On Aug 15, 2024, at 2:14 PM, Martin Husemann <[email protected]> wrote: > > On Thu, Aug 15, 2024 at 01:02:36PM -0400, vom513 wrote: >> So are you saying it?s possibly not actually crashing at that point and I should continue until I get something else / final crash ? > > Yes, exactly. When it stopy due to SIGILL check the call stack and just > "continue" at the gdb prompt if it is in a probe function. There will probably > a few instances of that very early, from a shared library that does > crypto stuff. > > Repeat untill you get to the real crash. Thanks Martin, this was the push I needed. I had thought the sparcv9 VIS probe stuff was it (red herring). For anyone that might see this in the future, ghostscript-agpl (10.x) seems to bus error on sparc due to color management engine alignment issues (lcms2): https://bugs.ghostscript.com/show_bug.cgi?id=695979#c20 I would guess it does this on other (non-x86) ports too. Someone was having similar issues a while back on HPUX. Adding "-dUseFastColor=true” to gs args uses the older color management engine and bypasses this error.