Follow-up Comment #9, bug #97585 (project xrootd):
Hi Brian,
Actually yesterday I was a bit too quick. The problem is not with comparing
two big numbers but with a multiplication (BN_mul) which gives a tiny, but
essential, difference in the two numbers to compare.
I made a few more checks comparing versions and I am more and more
convinced that the problem is with the 'bn' module of openssl.
Looking on the web it looks somehow known the 1.0.0 was rather buggy and
all major distributions, except RHEL, are already shipping with 1.0.1c since
months.
The last check I did was on my Ubuntu 12.04 machine where I can reproduce
the problem with 1.0.0j (the latest official 1.0.0x release). On such machine
I have created an hypothetical 1.0.0k by replacing by hand the 'crypto/bn'
module in 1.0.0j with the one coming from 1.0.1; the problem goes away.
I have done the same on lxplus6 (SLC6) and there too, the problem goes away
when I use the homemade '1.0.0k'.
Unfortunately the 'bn' model is quite cryptic and the changes between the
two versions are even more cryptic, so it is difficult to figure out what
exactly solves the problem.
I am not sure what is the best way to proceed. There are no workarounds
possible except removing the consistency check, which I do not want. I have
just started testing the ground with the SLC6 support for a version upgrade,
but I think I know what will be the first answers.
Gerri
Ps: By the way, I have been just told that Alice has also openssl-related
problems in running alien on SLC6, while they can run on Ubuntu 12.04 with
1.0.1c .
_______________________________________________________
Reply to this item at:
<http://savannah.cern.ch/bugs/?97585>
_______________________________________________
Message sent via/by LCG Savannah
http://savannah.cern.ch/
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1
