Good morning! I'm about to send the integration request and checking the mail_msg I'm seeing a diff in the ELF runtime attributes

17092a17093,17094

> usr/bin/adb: NEEDED=libc.so.1

> usr/bin/adb: RPATH=/usr/gcc/10/lib

31226,31227d31227

< usr/lib/isaexec: NEEDED=libc.so.1

< usr/lib/isaexec: RPATH=/usr/gcc/10/lib

I imagine this has nothing to do with the change for 15839 but I'm not entirely sure if it's got something to do with my setup, or is caused by other changes I've pulled in

tsoome: ping?

+Could not load the Python gdb module from `/usr/local/share/gdb/python'.

seem to have gotten past that now

+warning: could not convert 'breakme' from the host encoding (ISO-8859-1) to UTF-32.

+This normally should not happen, please file a bug report.

is that on omnios?

no openindiana hipster trying to get gdbserver working

kk

saw the latter on omnios, but andyf fixed that with guile update

I ran into that on omnios

yuripv: when was that fixed? In bloody, or is it in r151046 ?

in bloody, some days ago

(at least i'm running bloody)

DEECENT!

Okay, I'll have to give that a look

oh, paulf didn't realize you were in this channel too

I cannot reproduce what yuripv said about this being fixed in bloody.

andyf: does this issue ring any bells for you? That is, gdb fussing about "could not convert from host encoding to UTF"?

for what it's worth, I see the gdb encoding message on Solaris 11.4 as well

thanks alanc ! So, have any of y'all opened a gdb bug yet? I have an account on their bug tracker and can make one real quick

I haven't

paulf: have you?

looks like nothing related in the past few days. I am gonna go ahead and file this. Will drop a link here.

Rather than opening a new bug report, I just chimed in on sourceware.org/bugzilla/show_bug.cgi?id=29315

That one already mentions Solaris, and it's still marked as 'NEW'

pilonsi: didn't see anyone respond to you. that's a bit of non-determinism in pkgsend; it sees a file in the filesystem with N links into one "file" and N-1 "hardlink" actions and it sometimes chooses differently.

not opened any gdb bugs for that yet, just the problem with socketpair in the configure detection

pilonsi: excuse me, not pkgsend, but check_rtime

pkgsend also has that problem (though you can use --target to tell it which to pick), it's just not involved in this case

(ah, digging further it's usr/src/tools/find_elf/find_elf.c). it walks each directory in the order returned by readdir() rather than in a canonical order.

the original perl script did the same

the C version however (had to remember the specifics because it's been a while) does have special knowledge of certain 'well known' hardlinked binaries (usr/bin/alias and usr/lib/isaexec)

and tries to produce more 'expected' results when encountering hard links

if there's others where it makes sense, it'd be trivial to add

could also sort them, but if you add a new link that appears before the current first one, you end up in the same situation

[illumos-gate] 15487 smbios: storing the address of local variable -- Toomas Soome <tsoome⊙mc>

[illumos-gate] 11988 build DHCP client tools with CTF -- Joshua M. Clulow <josh⊙so>