Maybe I should throw away -mcpu and -march with --without-copt?
Or rather, replace them with an -mcpu that is appropriate for whatever it is that valgrind emulates.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2002-12-11 23:18: Subject: gcc 3.2
I don't see any good reason for that. Throw them away with --with-valgrind instead.
/ Martin Stjernholm, Roxen IS
My GMP seems to be compiled with MMX or more likely Athlon-XP instructions. Valgrind doesn't seem very easy to use on systems like mine...
| Testing vtables and scope access. | 1:2 2:6 3:20 4:73 5:280 6:1118 7:4590 8:19279 | disInstr: unhandled 2-byte opcode 0x6E | This _might_ be the result of executing an MMX, SSE, SSE2 or 3DNow! | instruction. Valgrind does not currently support such instructions. Sorry. | ==30970== | ==30970== Valgrind detected that your program requires | ==30970== the following unimplemented functionality: | ==30970== unhandled x86 0x0F 2-byte opcode | ==30970== This may be because the functionality is hard to implement, | ==30970== or because no reasonable program would behave this way, | ==30970== or because nobody has yet needed it. In any case, let me know | ==30970== (jseward@acm.org) and/or try to work around the problem, if you can. | ==30970== | ==30970== Valgrind has to exit now. Sorry. Bye! | ==30970== | | sched status: | | Thread 1: status = Runnable, associated_mx = 0x0, associated_cv = 0x0 | ==30970== at 0x2FB66560: __gmpn_preinv_divrem_1 (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FB6D895: __gmpn_get_str (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FB5CEC4: __gmpz_get_str (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FAE3723: low_get_mpz_digits (in /tmp/pikevalgrind/test-install/pike/7.5.1/lib/modules/Gmp.so) | | make: *** [verify] Error 1
/ Mirar
Previous text:
2002-12-12 09:24: Subject: gcc 3.2
I got it running with -mcpu=i486 -march=i486, so I'll check that into configure.in now. (pentium might work, but i486 definitely works.)
/ Mirar
Perhaps the easiest solution is to add these instructions to valgrind. But on the other hand, if it was easy they would probably have done so already...
/ Martin Nilsson (hehe Torgny)
Previous text:
2002-12-12 12:46: Subject: gcc 3.2
My GMP seems to be compiled with MMX or more likely Athlon-XP instructions. Valgrind doesn't seem very easy to use on systems like mine...
| Testing vtables and scope access. | 1:2 2:6 3:20 4:73 5:280 6:1118 7:4590 8:19279 | disInstr: unhandled 2-byte opcode 0x6E | This _might_ be the result of executing an MMX, SSE, SSE2 or 3DNow! | instruction. Valgrind does not currently support such instructions. Sorry. | ==30970== | ==30970== Valgrind detected that your program requires | ==30970== the following unimplemented functionality: | ==30970== unhandled x86 0x0F 2-byte opcode | ==30970== This may be because the functionality is hard to implement, | ==30970== or because no reasonable program would behave this way, | ==30970== or because nobody has yet needed it. In any case, let me know | ==30970== (jseward@acm.org) and/or try to work around the problem, if you can. | ==30970== | ==30970== Valgrind has to exit now. Sorry. Bye! | ==30970== | | sched status: | | Thread 1: status = Runnable, associated_mx = 0x0, associated_cv = 0x0 | ==30970== at 0x2FB66560: __gmpn_preinv_divrem_1 (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FB6D895: __gmpn_get_str (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FB5CEC4: __gmpz_get_str (in /usr/lib/libgmp.so.3.3.0) | ==30970== by 0x2FAE3723: low_get_mpz_digits (in /tmp/pikevalgrind/test-install/pike/7.5.1/lib/modules/Gmp.so) | | make: *** [verify] Error 1
/ Mirar
It's quite possibly they have already, in 1.1.0. Which on the other hand doesn't even start :)
/ Mirar
Previous text:
2002-12-12 12:53: Subject: gcc 3.2
Perhaps the easiest solution is to add these instructions to valgrind. But on the other hand, if it was easy they would probably have done so already...
/ Martin Nilsson (hehe Torgny)
Acording to the docs, it should be reasonably easy to add mmx support in the same way as the current fp-support.
/ Niels Möller ()
Previous text:
2002-12-12 12:53: Subject: gcc 3.2
Perhaps the easiest solution is to add these instructions to valgrind. But on the other hand, if it was easy they would probably have done so already...
/ Martin Nilsson (hehe Torgny)
It didn't help, though.
I can run some Pike stuff, --help works. --info doesn't, /that/ seems to be the NVidia problem (GTK?).
-e 'Image.Image' halts with | valgrind: the `impossible' happened: | unhandled x86 opcode
Hmm. More research tomorrow...
/ Mirar
Previous text:
2002-12-11 23:18: Subject: gcc 3.2
I don't see any good reason for that. Throw them away with --with-valgrind instead.
/ Martin Stjernholm, Roxen IS
pike-devel@lists.lysator.liu.se
-
Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum -
Martin Nilsson (hehe Torgny) ＠ Pike (-) developers forum
-
Martin Stjernholm, Roxen IS ＠ Pike developers forum
-
Mirar ＠ Pike developers forum
-
Niels M�ller () ＠ Pike (-) developers forum