On Oct 21 04:03, Andrew Schulman wrote: > > Instead, I found that COMSPEC needs to be set in the environment or I get a > > segfault as shown in my previous email below. I don't know why that is, > > but I can easily demonstrate it. > > Confirmed here. I had actually already seen this segfault too, but I > hadn't figured out it was caused by empty COMSPEC.
That's interesting. I just debugged this a bit. As you know, cygcheck is a native Windows application (it's supposed to work even if Cygwin is entirely broken). At one point it tries to fetch information about the installed services by calling cygrunsrv --list --verbose. It tries to accomplish that by calling the MSVCRT version of popen(2). And it's that call to popen which crashes if COMSPEC is not set, or not set correctly. I applied a patch to cygcheck, which enforces setting COMSPEC if the variable is unset. That doesn't help against *wrong* settings of COMSPEC, but that's really user's fault alone, I think :) I also created a new snapshot on https://cygwin.com/snapshots/ which comes with a cygcheck containing that patch. HTH, Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat
pgpiLrsecmM_I.pgp
Description: PGP signature