Re: [package - 144i386-default][textproc/xmlada] Failed for xmlada-26.0.0 in build/runaway

On 2026-07-09 12:48, Alastair Hogge wrote:
> On 2026-07-05 16:34, Alastair Hogge wrote:
>> On 2026-07-05 16:11, pkg-fallout@FreeBSD.org wrote:
>>> You are receiving this mail as a port that you maintain
>>> is failing to build on the FreeBSD package build server.
>>> Please investigate the failure and submit a PR to fix
>>> build.
>>> 
>>> Maintainer:     ada@FreeBSD.org
>>> Log URL:        https://pkg-status.freebsd.org/beefy18/data/144i386-default/2e12e7faaff2/logs/xmlada-26.0.0.log
>>> Build URL:      https://pkg-status.freebsd.org/beefy18/build.html?mastername=144i386-default&build=2e12e7faaff2
>>> Log:
>> 
>> [ remove bulk of the build log ]
>> 
>> 
>>> raised STORAGE_ERROR : s-intman.adb:136 explicit raise
>>> Call stack traceback locations:
>>> 0x85594c4
>>> 0x2887ff79
>>> 0x2887f417
>>> 0xffffea12
>>> 0x85c2b22
>>> 0x85c3d30
>>> 0x85af81d
>>> 0x85afa5c
>>> 0x85afbfc
>>> 0x85afcc1
>>> 0x85ce2f4
>>> 0x856912e
>>> 0x8568d0b
>>> 0x856956b
>>> 0x85695ea
>>> 0x8569315
>>> 0x85692f0
>>> 0x8101eca
>>> 0x85afcc1
>>> 0x85ce2f4
>>> 0x856912e
>>> 0x8568d0b
>>> 0x856956b
>>> 0x85695ea
>>> 0x8569315
>>> 0x85692f0
>>> 0x8101eca
>>> 0x85afcc1
>>> 0x85ce2f4
>>> 0x856912e
>>> 0x8568d0b
>>> 0x856956b
>>> 0x85695ea
>>> 0x8569315
>>> 0x85692f0
>>> 0x8101eca
>>> 0x85afcc1
>>> 0x85ce2f4
>>> 0x856912e
>>> 0x8568d0b
>>> 0x856956b
>>> 0x85695ea
>>> 0x8569315
>>> 0x85692f0
>>> 0x8101eca
>>> 0x85afcc1
>>> 0x85ce2f4
>>> 0x856912e
>>> 0x8568d0b
>>> 0x856956b
>>> =>> Killing runaway build after 10800 seconds with no output
>> 
>> These gprbuild problems on i386 surfaced 3 or more years ago. I noticed
>> themy when I was updating the de-orbited Ada Ports. I never found a
>> solution, other than maybe calling gnatmake directly instead of using
>> gprbuild. I have vague memories of modifying the environment (ulimit) to
>> increase memory, but I do not think that worked.
>> 
>> The Ada body s-intman (System.Interrupt_Management) is the same for both
>> 32 or 64 bit implementations. I suppose this is a good enuf reason to
>> try out the new build options for gprbuild.
> 
> So I have been using a poudriere-testport i386 interactive jail to try
> and understand this. The error is not with grpbuild, but gprconfig. I
> also noticed that even WITH_DEBUG devel/gdb claims there no debug
> symbols, even tho the binaries for gprbuild and gprconfig are ≅1MiB
> larger WITH_DEBUG, however, the bootstrap binaries in
> ${WRKDIR}/bootstrap/bin are ≅15MIB larger, and file(1) confirms that. So
> I copied the bootstrap binaries to ${PREFIX}/bin, and as expected, with
> debug symbols gprconfig does not crash. I tested a number of Ports, and
> they all built.
> 
> For DEBUG to work, I just added a new variable that is used for the
> build process in libgpr that attenuates the compiler warnings. I might
> submit a patch for that later.

I reverted to PRODUCTION builds of gprbuild, installed those binaries,
and ran make in textproc/xmlada again. This produces a core dump. I then
used gdb on the debug build of the gprbuild and the coredump:
# gdb
/wrkdirs/usr/ports/devel/gprbuild/work/gprbuild-26.0.0/exe/debug/gprbuild
/wrkdirs/usr/ports/textproc/xmlada/work/xmlada-26.0.0/gprbuild.core
(gdb) bt -ful
has been running for over one and half hours now.