On Saturday 22 April 2006 18:58, Russ Allbery wrote: > Bernd Schubert <[EMAIL PROTECTED]> writes: > > Package: krb5-admin-server > > Version: 1.4.3-6 > > Severity: important > > > > While testing kerberos on my home system I noticed that > > krb5-admin-server hangs on almost each system startup. After pressing > > randomly some keys the boot process goes on. While this is already > > annoying on my desktop system, it is not acceptable for a remotely > > administrated system. > > FWIW, I'm not seeing this on my systems, so I'm wondering what's different > between my setup and yours.
I'm surprised myself that I seem to be almost the only one to have this problem. However, one can some reports, e.g.: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=125743 > > > I think this should be fixable using the following patch and I'm just > > recompiling to test it. > > > > --- prng.c.old 2006-04-22 17:34:54.000000000 +0200 > > +++ prng.c 2006-04-22 17:35:21.000000000 +0200 > > @@ -164,7 +164,7 @@ > > int fd; > > unsigned char buf[YARROW_SLOW_THRESH/8]; > > int left; > > - fd = open (device, O_RDONLY); > > + fd = open (device, O_RDONLY | O_NONBLOCK); > > if (fd == -1) > > return 0; > > if (fstat (fd, &sb) == -1) > > It would surprise me if this helped at all, since if your understanding of > what's happening is correct, the hang is from trying to read entropy, not > just opening the device. Well, I would call read() 'subsequent operation', see man 2 open O_NONBLOCK or O_NDELAY When possible, the file is opened in non-blocking mode. Neither the open() nor any subsequent opera-tions on the file descriptor which is returned will cause the calling process to wait. For the han-dling of FIFOs (named pipes), see also fifo(4). This mode need not have any effect on files other than FIFOs. My unix programmers book from W. Richard Stevens also clearly states the non-blocking open mode is (also) for read and write operatations. Furthermore, after installing the recompiled packages and rebooting my system several time, I can confirm that the patch works. Cheers, Bernd -- Bernd Schubert PCI / Theoretische Chemie Universität Heidelberg INF 229 69120 Heidelberg
