John, Dan: Do either of you know about the issues from the original report?
I'm not familiar with the 64-bit issue for wireless-tools <29...Dan?
Only comment I see in the changelog about 64-bit stuff is:
* o Add temp variable to sscanf() to fix 64 bits issues [iwconfig]
I've backported this into the patch attached here, and the diff from
wireless-tools-29pre7 to wireless-tools-29pre8 from where I got the original
I'm a bit concerned about 'rts' and 'frag' sections. They both use temp
variables of 'long temp' but only 'rts' casts it to (unsigned long *) when
passing it to sscanf(). I'm not sure if that makes a difference or not, but I
checked in wireless-tools-29-pre10 (latest) and the 'rts' casting bits are still
John; can you take a look at see if there are any obvious mistakes? Can you
also compile on a 64-bit machine and check for warnings?
Created attachment 145922 [details]
Backport 64-bit sscanf argument fixes from wireless-tools-29-pre8 to wireless-tools-28
Created attachment 145923 [details]
Diff from wt-29-pre7 to wt-29-pre8 in which sscanf 64-bit fixes were added upstream
Basically, without a compelling reason, clear feature request, or a bunch of
regression testing, I wouldn't feel comfortable with just revving wireless-tools
up to v29. I feel like most issues could be resolved fairly easily through
backports since the code is pretty contained and fairly straightforward.
I agree with the last comment. Please backport. I'll change the summary.
Building as wireless-tools-28-2.el5
Just wanted to note that we can't fully test this fix for regressions because of
No problems. The new wireless-tools don't appear to break anything.
wireless-tools-28-2.el5 included in 20070124.1.