Hi,
I am running a multi-threaded (64 threads) simulation in x86 SE mode. I am
encountering some issues which (as far as my current analysis goes) are to
do with libc support in Gem5. I have been seeing errors of the nature -

1)panic: Tried to read unmapped address 0xxx
2)glibc deleted ** double free or corruption (fasttop)

I read that these are basically segfaults in the source application, so I
tried producing a stack-trace and found that the error come mainly from the
libc library. As an example, I had a push_back() to a vector in my code
which was calling the 'new' operator to resize the vector. A fault was
coming from inside this 'new operator' call. Changing the code to eliminate
the push_back() solved that error. But I am seeing this at other places
too, the stack-trace being -

[0] _int_malloc
[1] <__libc_malloc>
[2] <operator new(unsigned long)>
[3] ...

Have people encountered problems like this before?

I am using libc 2.12 and c++11.

Thanks.
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to