Suspected bug in longjmp()

[Paul Sokolovsky]

Hello cygwin32-developers@cygnus.com,

  In my sources of b20.1 release, I see that longjmp() restores %fs
register. This might not be done, because %fs is system (in sence
'belonging to system') register in win32, and longjmp()'s work is to
resore application's state, not system's.

  Of course, this note comes not from speculative conclusion, but from
my practise with implementing my own fork() - on child's exit I
sometimes (about every tenth time) got exception in win95 kernel32.dll
on usual 'mov eax,fs:[0]' stuff: fs contained null selector. Not
restoring fs helped.

Best regards,
 Paul                          mailto:paul-ml@is.lg.ua



--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com