http://bugs.winehq.org/show_bug.cgi?id=58335
--- Comment #64 from Bernhard Übelacker bernhardu@mailbox.org --- Created attachment 78880 --> http://bugs.winehq.org/attachment.cgi?id=78880 gdb-commands-58335_6.txt
(In reply to Michael Setzer II from comment #63)
Created attachment 78878 [details] output 5
output of 5.txt
Thank you very much, this output shows indeed the memory got written a zero to. And there is also a movaps instruction before which copies the content from register xmm6 into the memory which should hold the pointer we need (and which it does in my VM).
So this moves the AVX thing back into focus I guess ...
Hardware watchpoint 7: *(0x7ffffe2ff7e0) Old value = -30257520 New value = 0
0x6fffffc0cfd0 <+912>: movdqa 0xf0(%rsp),%xmm6 # line 5378 66 0f 6f b4 24 f0 00 00 00 ... 0x6fffffc0d9ae <+3438>: movaps %xmm6,0xf0(%rsp) # line 3280 0f 29 b4 24 f0 00 00 00 => 0x6fffffc0d9b6 <+3446>: mov %rax,0x28(%rsp) # line 3280
I had guessed there would be an exception when unsupported instructions get executed.
Just to be sure, which value is in this xmm6 register and when, could you please execute this again: wineserver -k script -c "gdb -q < gdb-commands-58335_6.txt" -a gdb-output-58335_6.txt