If switching from SGI as to GNU as 2.15, a couple of new testsuite failures occur on the 3.4 branch as of 20050607: +FAIL: gcc.dg/range-test-1.c (test for excess errors) FAIL: gcc.dg/range-test-1.c (test for excess errors) Excess errors: /var/tmp//ccj9FSrW.s:1483: Warning: .space repeat count is zero, ignored /var/tmp//ccj9FSrW.s:1506: Warning: .space repeat count is zero, ignored A couple of libjava failures have the same cause: +FAIL: PR295 compilation from source +FAIL: PR295 compilation from bytecode +FAIL: PR295 -O3 compilation from source +FAIL: PR295 -O3 compilation from bytecode +FAIL: Statics compilation from source +FAIL: Statics compilation from bytecode +FAIL: Statics -O3 compilation from source +FAIL: Statics -O3 compilation from bytecode @@ -255,0 +266,4 @@ +FAIL: Class_1 compilation from source +FAIL: Class_1 compilation from bytecode +FAIL: Class_1 -O3 compilation from source +FAIL: Class_1 -O3 compilation from bytecode @@ -259,0 +274,8 @@ +FAIL: PR12416 compilation from source +FAIL: PR12416 compilation from bytecode +FAIL: PR12416 -O3 compilation from source +FAIL: PR12416 -O3 compilation from bytecode +FAIL: PR7482 compilation from source +FAIL: PR7482 compilation from bytecode +FAIL: PR7482 -O3 compilation from source +FAIL: PR7482 -O3 compilation from bytecode +FAIL: pr179 compilation from source +FAIL: pr179 compilation from bytecode +FAIL: pr179 -O3 compilation from source +FAIL: pr179 -O3 compilation from bytecode +FAIL: verify compilation from source +FAIL: verify compilation from bytecode +FAIL: verify -O3 compilation from source +FAIL: verify -O3 compilation from bytecode .space 0 is generated for both assemblers, but while SGIs MIPSpro as silently ignores it, gas warns as above. Environment: System: IRIX sculptor 6.5 10060437 IP32 host: mips-sgi-irix6.5 build: mips-sgi-irix6.5 target: mips-sgi-irix6.5 configured with: /vol/gnu/src/gcc/gcc-3.4-branch-dist/configure --prefix=/vol/gcc --with-local-prefix=/vol/gcc --disable-nls --with-gnu-as --with-as=/vol/gcc/lib/gas-2.15 --enable-libgcj --disable-multilib How-To-Repeat: Compare testsuite results with and without gas.
Did you happen to catch where the .space 0 were being emitted from?
Subject: Re: New IRIX 6.5 testsuite failures with gas: .space repeat count is zero, ignored echristo at redhat dot com writes: > Did you happen to catch where the .space 0 were being emitted from? Not yet: I've just posted the bug report so the issue isn't forgotten. Rainer
Subject: Re: New IRIX 6.5 testsuite failures with gas: .space repeat count is zero, ignored echristo at redhat dot com writes: > Did you happen to catch where the .space 0 were being emitted from? I've now managed to distill a test case: int test (void) { static unsigned long noarro1 [] = { }; } % ../xgcc -B../ -c space0.i /var/tmp//cc3uQnef.s: Assembler messages: /var/tmp//cc3uQnef.s:8: Warning: .space repeat count is zero, ignored Rainer
struct A {} a; is another testcase.
Is this still an issue?
I haven't seen this in a long time, and the testcases pass since they now use .comm instead of .space. I meant to keep the PR open to investigate why gcc would emit code like this, but I think it can now be closed, especially given that IRIX support will be removed in 4.8. Rainer
4.7 branch closed.