you may want to try running your program with valgrind and seeing if there are any memory leaks that are causing nondeterministic results.
On Fri, Oct 7, 2011 at 8:30 PM, Rio Xiangyu Dong <[email protected]>wrote: > I am having a page table fault issue when running a benchmark on ARM_SE. > The error message I got was:**** > > ** ** > > […]**** > > warn: mremapping to totally new vaddr 0x416ce000-0x41746000, adding 491520 > **** > > warn: returning 0x416ce000 as start**** > > warn: mremapping to totally new vaddr 0x41746000-0x4193c000, adding 2056192 > **** > > warn: returning 0x41746000 as start**** > > warn: Not extending stack: address 0x41421758 isn't at the end of the > stack.**** > > panic: Page table fault when accessing virtual address 0x41421758**** > > @ cycle 786760171000**** > > [invoke:build/ARM_SE/sim/faults.cc, line 70]**** > > Memory Usage: 2197968 KBytes**** > > Program aborted at cycle 786760171000**** > > Abort (core dumped)**** > > ** ** > > It seems to me that the address 0x41421758 is not in the emulated page > table. However, the log shows “mremapping” adds the memory space up to > 0x4193c000 already. Also, the most interesting thing is: this issue CANNOT > be reproduced on another almost identical machine (I mean it can run > successfully on another machine as follows):**** > > ** ** > > […]**** > > warn: mremapping to totally new vaddr 0x416ce000-0x41746000, adding 491520 > **** > > warn: returning 0x416ce000 as start**** > > warn: mremapping to totally new vaddr 0x41746000-0x4193c000, adding 2056192 > **** > > warn: returning 0x41746000 as start**** > > warn: mremapping to totally new vaddr 0x4193c000-0x41a29000, adding 970752 > **** > > warn: returning 0x4193c000 as start**** > > warn: mremapping to totally new vaddr 0x41a41000-0x41ad1000, adding 589824 > **** > > warn: returning 0x41a41000 as start**** > > […]**** > > [looks like it’s going to run forever]**** > > ** ** > > **1. **two machines (both linux 2.6 on x86-64), the failed one has > larger memory size.**** > > **2. **I have identical source codes on these two machines**** > > **3. **I compile the source codes cleanly**** > > **4. **I set the same “ulimit”**** > > **5. **The only difference I can see is the compiler toolchain: one > uses GCC 4.4.5/Python 2.6.7, one uses GCC4.4.6/Python 2.6.6**** > > ** ** > > I would appreciate it very much if someone can give me some hints.**** > > ** ** > > Thank you!**** > > _______________________________________________ > gem5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users > -- - Korey
_______________________________________________ gem5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
