sysconf can call malloc?

[Roland McGrath]

I don't disagree with your reasoning about the complexity of things sysconf
has to do.  It seems entirely reasonable to me that applications and
libraries get the info they need from sysconf at startup time when it's
easy to be in a safe place.

But, POSIX.1 does list it as async-signal-safe.  Any library writer, no
matter what kind of library they are writing, is completely reasonable to
expect it to be async-signal-safe given today's standard.

Are you working to change that specification?


Thanks,
Roland