Re: Memory handling bug in 0.9.8a AES assembler code for x86 ?

2006-04-21 Thread Andy Polyakov
When I run the the openssl evptest suite in purify runtime memory usage analysis tool it reports lots of memory segment errors in the AES crypto functions. Question is how does purify calculate this segment boundary. Obviously it doesn't look at current %esp value [as there are no references

Re: Memory handling bug in 0.9.8a AES assembler code for x86 ?

2006-04-21 Thread Andy Polyakov
Is it possible to make purify point out offending instruction? Maybe it's possible to make it dump the code instead of reporting error, in which case you can use debugger to find the offending instruction... Should read dump the core, not code. A.

[openssl.org #1320] Memory handling bug in 0.9.8a AES assembler code for x86

2006-04-20 Thread Leif Thuresson via RT
When I run the the openssl evptest suite in purify runtime memory usage analysis tool it reports lots of memory segment errors in the AES crypto functions. If I build openssl with AES assembler code disabled (e.i AES in c-code) the same test runs without any errors from purify. The test was

Memory handling bug in 0.9.8a AES assembler code for x86 ?

2006-04-13 Thread Leif Thuresson
When I run the the openssl evptest suite in purify runtime memory usage analysis tool it reports lots of memory segment errors in the AES crypto functions. If I build openssl with AES assembler code disabled (e.i AES in c-code) the same test runs without any errors from purify. Anyone seen