Jeffrey Altman <[EMAIL PROTECTED]> wrote:
Tom Keiser wrote:
On 2/14/06, Derrick J Brashear <[EMAIL PROTECTED]> wrote:
Did you --enable-fast-restart (I think that's what it is)? Did you
do that also when you did 1.4.1-rc6? Basically I'm wondering if the
CVS head is really broken or if you had a mixed set of binaries in
play.
I can confirm that we have a problem. Basically, the windows-64 type
safety patch changed a bunch of variable declarations of type long to
type size_t, which changed their signedness. This broke a few
comparison operator checks in the lwp iomgr, resulting in
IOMGR_Select returning 1 instead of -1. Thus, bproc() never reaps
its children, and the bnodes never change state. I don't really
have time right now to work on a patch, but that's the source of the
bosserver problems.
Regards,
-Tom
Ah yes. The joy of using a variable to store both numeric values
under some circumstances and pointers at other times.
Does this mean I'm not crazy?
This patch will fix the problem. Sorry for the inconvenience.
In what instances should this patch be applied? And to what source?
rc7? cvs head? cvs stable?
<<CDC
_______________________________________________
OpenAFS-devel mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-devel
