With HEAD binutils checked out as of Wed, 27 Dec 2006 06:05:09 UTC, I get these failues for "make check-Y" for --target=d10v-elf: Running /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/lns/lns.exp ... FAIL: lns-common-1 ... Running /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/macros/macros.exp ... FAIL: macros dot FAIL: macros purge with the gas.log file containing: /mnt/test1/hp/axis/binutils/d10v/gas/testsuite/../../binutils/readelf -wl dump.o > dump.out extra regexps in /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/lns/lns-common-1.d starting with "^Dump of debug contents of s\ ection \.debug_line:$" EOF from dump.out FAIL: lns-common-1 ... Executing /mnt/test1/hp/axis/binutils/src/gas/testsuite/lib/run ../as-new -alm /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas\ /macros/dot.s >&dump.out regexp_diff match failure regexp "^.*:28: Error: .*$" line "/mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/macros/dot.s:28: Fatal error: unknown opcode: x.y.z" extra regexps in /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/macros/dot.l starting with "^[ ]*[1-9][0-9]*[ ]+m 4, 2$\ " EOF from dump.out FAIL: macros dot ... Executing /mnt/test1/hp/axis/binutils/src/gas/testsuite/lib/run ../as-new --hash-size=8000 /mnt/test1/hp/axis/binutils/src/gas/t\ estsuite/gas/macros/purge.s >&dump.out regexp_diff match failure regexp "^.*:12: Error: .*$" line "/mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/macros/purge.s:12: Fatal error: unknown opcode: macro1" extra regexps in /mnt/test1/hp/axis/binutils/src/gas/testsuite/gas/macros/purge.l starting with "^.*:13: Error: .*$" EOF from dump.out FAIL: macros purge
Hi Hans-Peter, The d10v port should now have no GAS testsuite failures. There were two problems. The first is that the d10v port does not produce dwarf2 line number information. This is because the port was never upgraded to do so, and since nobody has complained about it I see no reason to add the support now. Instead I have just disabled the line nubmer test for the d10v port. The other problem was that the port would stop with a fatal error whenever it encountered an unrecognizable opcode, rather than just issuing a normal error message and carrying on. This meant that a couple of the macro tests did not work. Stopping assembly at the first unknown opcode seems a bit harsh to me, so I have changed the d10v port to emit an error message and continue. Cheers Nick gas/ChangeLog 2007-02-28 Nick Clifton <nickc@redhat.com> PR gas/3797 * config/tc-d10v.c (do_assemble): Do not generate error messages, just return -1 whenever a problem is encountered. (md_assemble): If do_assemble returns -1 generate a non-fatal error message and return. gas/testsuite/ChangeLog 2007-02-28 Nick Clifton <nickc@redhat.com> PR gas/3797 * gas/lns/lns.exp: Do not run the lns-common test for the d10v port. * gas/d10v/address-002.l: Update expected assembler output. * gas/d10v/address-003.l, gas/d10v/address-004.l, gas/d10v/address-005.l, gas/d10v/address-006.l, gas/d10v/address-007.l, gas/d10v/address-008.l, gas/d10v/address-009.l, gas/d10v/address-010.l, gas/d10v/address-011.l, gas/d10v/address-012.l, gas/d10v/address-013.l, gas/d10v/address-014.l, gas/d10v/address-015.l, gas/d10v/address-016.l, gas/d10v/address-017.l, gas/d10v/address-018.l, gas/d10v/address-019.l, gas/d10v/address-020.l, gas/d10v/address-021.l, gas/d10v/address-022.l, gas/d10v/address-023.l, gas/d10v/address-024.l, gas/d10v/address-025.l, gas/d10v/address-026.l, gas/d10v/address-027.l, gas/d10v/address-030.l, gas/d10v/address-031.l, gas/d10v/address-032.l, gas/d10v/address-033.l, gas/d10v/address-034.l, gas/d10v/address-035.l, gas/d10v/address-036.l, gas/d10v/address-037.l, gas/d10v/address-038.l, gas/d10v/address-039.l, gas/d10v/address-040.l, gas/d10v/address-041.l: Likewise.