> On Tue, 15 Sep 2009, Gabriel Michael Black wrote: > As promised here's a patch. It turns out I was already attempting to > sign extend the second operand (that's what sop2 is supposed to be > doing), I just dropped the ball for the immediate variant.
Unfortunately this patch doesn't seem to fix things for me, it still fails in the same place. Details from the trace: 263500: system.cpu T0 : @store_byte+13.0 : LOOP_I : rdip t1d, %ctrl153, : IntAlu : D=0x0000000008048098 264000: system.cpu T0 : @store_byte+13.1 : LOOP_I : subi ecx, ecx, 0x1 : IntAlu : D=0x0000000000000000 264500: system.cpu T0 : @store_byte+13.2 : LOOP_I : wripi , t1d, 0xe7 : IntAlu : 265000: system.cpu T0 : @print_s+34.0 : PUSH_R : stupd esi, SS:[esp + 0xfffffffffffffffc] : MemWrite : D=0x00000000ffffdf18 A=0xffffdf18 The code in question looks like: 0804807f <output_loop>: ... 08048089 <store_byte>: ... 8048096: e2 e7 loop 804807 ... 0804815d <print_s>: ... 804817f: 56 push %esi As you can see, the loop instruction, instead of looping to 8048098+ffffffe7 = 804807f is looping to 8048098+000000e7 = 804817f The test suite I am using can be found at: http://www.deater.net/weave/vmwprod/asm/ll/qemu_tests.html Download, uncompress, and run "test_m5.sh" in the ./qemu_test directory after setting up the script to point to the location of m5 on your machine. Vince _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
