igork loader starting address is printed out quite early; however, sine it was dboot blowing up, I'd start with dboot and its address is recorded in unix elf header - 0xc00000 (the PT_LOAD entry with vaddr and paddr having the same value.

I have a question regarding our binutils. While building the latest 2.42 version I found out that OI has one patch that OmniOS doesn't have, and OmniOS has two additional patches that OI doesn't have. Solaris-userland has only one patch (for binutils-2.40) that neither OI nor OmniOS have.

The patches OI and OmniOS have in common are named (OmniOS) ld64so.patch andld64so.patch

The patches only OmniOS has are gold-fallocate.patch and startstop.patch.

Despite startstop.patch no patch comes with an explanation what makes it hard to check whether they are needed or not.

startstop.patch reverts a commit from upstream that is more than 3 years old.

I wonder why OI hasn't been hit by the problem it should resolve yet.

Solaris userland has a patch named 00_gstrip.patch which removes .SUNW_sym.*sort sections according to its comment.

OI has a patch named 02-gas.write.c.patch which seems to be quite old (2012?) nobody else seem to use.

My question is: which of these patches are really needed? Does anybody have an idea?

posix_fallocate looks like it's to avoid C++ namespace damage, ld64.so is so gold uses the correct ELF interpretter, stdio-limit explains itself in the patch, it's to try to get more FILE's on 32bit systems, or limit itself to not cross the limit if it fails.

all seem necessary except perhaps posix_fallocate, if that got fixed

startstop might be unnecessary if upstream fixed the referenced bug

brb

upstream commented the bug report: the change was intentional and will not be reverted.

Thanks for the explanations. I will add them to the patches (your ok assumed).

you'd want to ask toomas about how FreeBSD fixed that, I guess

and yeah, add those if it helps :)

Thanks again.

freebsd does not depend on gnu binutils.