We got a report from a customer that their ThinLinc Client 4.9.0 crashed when they tried to use public key authentication. They kindly included an backtrace which identified the problem. > Program terminated with signal SIGSEGV, Segmentation fault. > #0 __strlen_avx2 () at ../sysdeps/x86_64/multiarch/strlen-avx2.S:62 > 62 ../sysdeps/x86_64/multiarch/strlen-avx2.S: Aucun fichier ou dossier de ce type. > (gdb) where > #0 __strlen_avx2 () at ../sysdeps/x86_64/multiarch/strlen-avx2.S:62 > #1 0x00007f81395034d3 in _IO_vfprintf_internal (s=s@entry<mailto:s=s@entry>=0x7fffed8a98d0, format=format@entry<mailto:format=format@entry>=0x7f813a010db7 <error: Cannot access memory at address 0x7f813a010db7>, > ap=ap@entry<mailto:ap=ap@entry>=0x7fffed8a9ef8) at vfprintf.c:1643 > #2 0x00007f813952e910 in _IO_vsnprintf (string=0x7fffed8a9a80 "Phrase de chiffrement (PIN) du symbole \"", maxlen=<optimized out>, > format=0x7f813a010db7 <error: Cannot access memory at address 0x7f813a010db7>, args=0x7fffed8a9ef8) at vsnprintf.c:114 > ...
> * tlclient should not crash when logging in with public key > authentication using a french locale I can see the crash with 4.9.0, but not with trunk. It also works fine with translations that didn't have this issue. > * There should be no format string errors in any of our > translations, neither tlclient nor tl-misc. I can not find any instance where things are wrong. > * msgfmt should fail to build mo files from po files with > format string errors Seems to work. I tested both changing the number of specifiers (adding or removing), and changing the type. It complained about both. It does not complain about changes in precision, but I guess that is okay. Note that msgfmt is a bit naughty and leaves the generated .mo file in place. So if you run make again you'll get a success. Let's see if that can be fixed.
Rules fixed now. And release notes look good. That should be everything.