* The new ia64 assembler is incompatible with glibc
@ 2005-02-14 6:32 H. J. Lu
2005-02-14 8:35 ` H. J. Lu
0 siblings, 1 reply; 4+ messages in thread
From: H. J. Lu @ 2005-02-14 6:32 UTC (permalink / raw)
To: jbeulich; +Cc: wilson, binutils
[-- Attachment #1: Type: text/plain, Size: 1086 bytes --]
After applied 2 patches enclosed here, I can compile glibc in CVS. But
I got so many failures in "make check":
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/basic-test.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-double.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-float.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-idouble.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ifloat.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ldouble.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ildoubl.out]
Error 1
make[3]: *** [math/xtests] Error 2
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstscanf.out]
Error 1
make[4]: ***
[/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstdiomisc.out]
Error 1
It looks like the math functions are wrong. Replacing the new assembler
with the old one fixes the problem.
H.J.
[-- Attachment #2: libc-ia64-unwind-1.patch --]
[-- Type: text/plain, Size: 673 bytes --]
--- sysdeps/ia64/elf/initfini.c.unwind 2004-08-18 16:13:30.000000000 -0700
+++ sysdeps/ia64/elf/initfini.c 2005-02-03 15:53:18.000000000 -0800
@@ -116,13 +116,14 @@ __asm__ (".section .init\n"
"/*@_init_PROLOG_ENDS*/\n"
"\n"
"/*@_init_EPILOG_BEGINS*/\n"
+" .section .init\n"
" .proc _init#\n"
+"_init:\n"
" .prologue\n"
" .save ar.pfs, r34\n"
" .vframe r32\n"
" .save rp, r33\n"
" .body\n"
-" .section .init\n"
" .regstk 0,2,0,0\n"
" mov r12 = r32\n"
" mov ar.pfs = r34\n"
@@ -155,6 +156,7 @@ __asm__ (".section .init\n"
"/*@_fini_EPILOG_BEGINS*/\n"
" .section .fini\n"
" .proc _fini#\n"
+"_fini:\n"
" .prologue\n"
" .save ar.pfs, r34\n"
" .vframe r32\n"
[-- Attachment #3: libc-ia64-munwind-1.patch --]
[-- Type: text/plain, Size: 3952 bytes --]
2005-02-10 H.J. Lu <hongjiu.lu@intel.com>
* sysdeps/ia64/fpu/e_scalb.S: Replace __libm_error_region: with
LOCAL_LIBM_ENTRY(__libm_error_region)
* sysdeps/ia64/fpu/e_scalbf.S: Likewise
* sysdeps/ia64/fpu/e_scalbl.S: Likewise
* sysdeps/ia64/fpu/libm_scalblnf.S: Likewise
* sysdeps/ia64/fpu/s_libm_ldexp.S: Likewise
* sysdeps/ia64/fpu/s_libm_ldexpf.S: Likewise
* sysdeps/ia64/fpu/s_libm_ldexpl.S: Likewise
* sysdeps/ia64/fpu/s_libm_scalbn.S: Likewise
* sysdeps/ia64/fpu/s_libm_scalbnf.S: Likewise
* sysdeps/ia64/fpu/s_libm_scalbnl.S: Likewise
--- sysdeps/ia64/fpu/e_scalb.S.munwind 2005-01-07 14:13:58.000000000 -0800
+++ sysdeps/ia64/fpu/e_scalb.S 2005-02-03 16:00:20.000000000 -0800
@@ -461,7 +461,7 @@ SCALB_NAN_INF_ZERO:
};;
GLOBAL_IEEE754_END(scalb)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALB_OVERFLOW:
SCALB_UNDERFLOW:
--- sysdeps/ia64/fpu/e_scalbf.S.munwind 2005-01-07 14:13:58.000000000 -0800
+++ sysdeps/ia64/fpu/e_scalbf.S 2005-02-03 16:00:29.000000000 -0800
@@ -461,7 +461,7 @@ SCALBF_NAN_INF_ZERO:
};;
GLOBAL_IEEE754_END(scalbf)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALBF_OVERFLOW:
SCALBF_UNDERFLOW:
--- sysdeps/ia64/fpu/e_scalbl.S.munwind 2005-01-07 14:13:58.000000000 -0800
+++ sysdeps/ia64/fpu/e_scalbl.S 2005-02-03 16:00:41.000000000 -0800
@@ -461,7 +461,7 @@ SCALBL_NAN_INF_ZERO:
};;
GLOBAL_IEEE754_END(scalbl)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALBL_OVERFLOW:
SCALBL_UNDERFLOW:
--- sysdeps/ia64/fpu/libm_scalblnf.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/libm_scalblnf.S 2005-02-03 16:00:45.000000000 -0800
@@ -296,7 +296,7 @@ GLOBAL_LIBM_ENTRY(__libm_scalblnf)
}
GLOBAL_LIBM_END(__libm_scalblnf)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
scalbnf_OVERFLOW:
scalbnf_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_ldexp.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_ldexp.S 2005-02-03 15:56:52.000000000 -0800
@@ -296,7 +296,7 @@ GLOBAL_LIBM_ENTRY(__libm_ldexp)
}
GLOBAL_LIBM_END(__libm_ldexp)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
LDEXP_OVERFLOW:
LDEXP_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_ldexpf.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_ldexpf.S 2005-02-03 15:57:07.000000000 -0800
@@ -294,7 +294,7 @@ GLOBAL_LIBM_ENTRY(__libm_ldexpf)
}
GLOBAL_LIBM_END(__libm_ldexpf)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
LDEXPF_OVERFLOW:
LDEXPF_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_ldexpl.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_ldexpl.S 2005-02-03 15:55:21.000000000 -0800
@@ -295,7 +295,7 @@ GLOBAL_LIBM_ENTRY(__libm_ldexpl)
}
GLOBAL_LIBM_END(__libm_ldexpl)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
LDEXPL_OVERFLOW:
LDEXPL_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_scalbn.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_scalbn.S 2005-02-03 16:00:49.000000000 -0800
@@ -295,7 +295,7 @@ GLOBAL_LIBM_ENTRY(__libm_scalbn)
}
GLOBAL_LIBM_END(__libm_scalbn)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALBN_OVERFLOW:
SCALBN_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_scalbnf.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_scalbnf.S 2005-02-03 16:00:53.000000000 -0800
@@ -295,7 +295,7 @@ GLOBAL_LIBM_ENTRY(__libm_scalbnf)
}
GLOBAL_LIBM_END(__libm_scalbnf)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALBNF_OVERFLOW:
SCALBNF_UNDERFLOW:
--- sysdeps/ia64/fpu/s_libm_scalbnl.S.munwind 2005-01-06 03:29:22.000000000 -0800
+++ sysdeps/ia64/fpu/s_libm_scalbnl.S 2005-02-03 16:01:07.000000000 -0800
@@ -295,7 +295,7 @@ GLOBAL_LIBM_ENTRY(__libm_scalbnl)
}
GLOBAL_LIBM_END(__libm_scalbnl)
-__libm_error_region:
+LOCAL_LIBM_ENTRY(__libm_error_region)
SCALBNL_OVERFLOW:
SCALBNL_UNDERFLOW:
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: The new ia64 assembler is incompatible with glibc
2005-02-14 6:32 The new ia64 assembler is incompatible with glibc H. J. Lu
@ 2005-02-14 8:35 ` H. J. Lu
2005-02-14 9:36 ` PATCH: Add an ia64 gas testcase for @nat|@qnan H. J. Lu
0 siblings, 1 reply; 4+ messages in thread
From: H. J. Lu @ 2005-02-14 8:35 UTC (permalink / raw)
To: jbeulich; +Cc: wilson, binutils
On Sun, Feb 13, 2005 at 09:18:34AM -0800, H. J. Lu wrote:
> After applied 2 patches enclosed here, I can compile glibc in CVS. But
> I got so many failures in "make check":
>
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/basic-test.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-double.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-float.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-idouble.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ifloat.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ldouble.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ildoubl.out]
> Error 1
> make[3]: *** [math/xtests] Error 2
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstscanf.out]
> Error 1
> make[4]: ***
> [/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstdiomisc.out]
> Error 1
>
> It looks like the math functions are wrong. Replacing the new assembler
> with the old one fixes the problem.
>
The problem is caused by
http://sourceware.org/ml/binutils/2005-02/msg00122.html
Gas no longer handles
fclass.m p6, p7 = farg0, @inf|@pos
correctly. Jan, could you please fix it?
Thanks.
H.J.
^ permalink raw reply [flat|nested] 4+ messages in thread
* PATCH: Add an ia64 gas testcase for @nat|@qnan
2005-02-14 8:35 ` H. J. Lu
@ 2005-02-14 9:36 ` H. J. Lu
2005-02-15 0:00 ` James E Wilson
0 siblings, 1 reply; 4+ messages in thread
From: H. J. Lu @ 2005-02-14 9:36 UTC (permalink / raw)
To: jbeulich; +Cc: wilson, binutils
On Sun, Feb 13, 2005 at 10:42:15AM -0800, H. J. Lu wrote:
> On Sun, Feb 13, 2005 at 09:18:34AM -0800, H. J. Lu wrote:
> > After applied 2 patches enclosed here, I can compile glibc in CVS. But
> > I got so many failures in "make check":
> >
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/basic-test.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-double.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-float.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-idouble.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ifloat.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ldouble.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/math/test-ildoubl.out]
> > Error 1
> > make[3]: *** [math/xtests] Error 2
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstscanf.out]
> > Error 1
> > make[4]: ***
> > [/export/build/gnu/glibc-nptl/build-ia64-linux/stdio-common/tstdiomisc.out]
> > Error 1
> >
> > It looks like the math functions are wrong. Replacing the new assembler
> > with the old one fixes the problem.
> >
>
> The problem is caused by
>
> http://sourceware.org/ml/binutils/2005-02/msg00122.html
>
> Gas no longer handles
>
> fclass.m p6, p7 = farg0, @inf|@pos
>
> correctly. Jan, could you please fix it?
>
I am going to check in this testcase to catch the error.
H.J.
----
2005-02-13 H.J. Lu <hongjiu.lu@intel.com>
* gas/ia64/ia64.exp: Add "operand-or".
* gas/ia64/operand-or.d: New file.
* gas/ia64/operand-or.s: Likewise.
--- gas/testsuite/gas/ia64/ia64.exp.or 2005-02-11 13:21:20.000000000 -0800
+++ gas/testsuite/gas/ia64/ia64.exp 2005-02-13 11:18:23.604690961 -0800
@@ -72,4 +72,6 @@ if [istarget "ia64-*"] then {
run_list_test "unwind-err" "-munwind-check=error"
run_list_test "hint.b-err" ""
run_list_test "hint.b-warn" "-mhint.b=warning"
+
+ run_dump_test "operand-or"
}
--- gas/testsuite/gas/ia64/operand-or.d.or 2005-02-13 11:18:23.604690961 -0800
+++ gas/testsuite/gas/ia64/operand-or.d 2005-02-13 11:18:23.604690961 -0800
@@ -0,0 +1,29 @@
+# objdump: -d --disassemble-zeroes
+# name: ia64 operand-or
+
+.*: +file format .*
+
+Disassembly of section \.text:
+
+0+000 <_start>:
+ 0: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 6: 30 20 80 09 28 00 fclass\.m p3,p4=f4,0x180
+ c: 00 00 00 20 nop\.b 0x0
+ 10: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 16: 30 20 c0 09 28 00 fclass\.m p3,p4=f4,0x1c0
+ 1c: 00 00 00 20 nop\.b 0x0
+ 20: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 26: 30 20 c0 89 28 00 fclass\.m p3,p4=f4,0x1c1
+ 2c: 00 00 00 20 nop\.b 0x0
+ 30: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 36: 30 20 c0 89 29 00 fclass\.m p3,p4=f4,0x1c3
+ 3c: 00 00 00 20 nop\.b 0x0
+ 40: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 46: 30 20 c8 89 29 00 fclass\.m p3,p4=f4,0x1cb
+ 4c: 00 00 00 20 nop\.b 0x0
+ 50: 1c 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 56: 30 20 d8 89 29 00 fclass\.m p3,p4=f4,0x1db
+ 5c: 00 00 00 20 nop\.b 0x0
+ 60: 1d 00 00 00 01 00 \[MFB\] nop\.m 0x0
+ 66: 30 20 f8 89 29 00 fclass\.m p3,p4=f4,0x1fb
+ 6c: 00 00 00 20 nop\.b 0x0;;
--- gas/testsuite/gas/ia64/operand-or.s.or 2005-02-13 11:18:23.605690831 -0800
+++ gas/testsuite/gas/ia64/operand-or.s 2005-02-13 11:18:23.605690831 -0800
@@ -0,0 +1,11 @@
+.text
+ .type _start,@function
+_start:
+
+ fclass.m p3, p4 = f4, @nat|@qnan
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan|@pos
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan|@pos|@neg
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan|@pos|@neg|@unorm
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan|@pos|@neg|@unorm|@norm
+ fclass.m p3, p4 = f4, @nat|@qnan|@snan|@pos|@neg|@unorm|@norm|@inf
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: PATCH: Add an ia64 gas testcase for @nat|@qnan
2005-02-14 9:36 ` PATCH: Add an ia64 gas testcase for @nat|@qnan H. J. Lu
@ 2005-02-15 0:00 ` James E Wilson
0 siblings, 0 replies; 4+ messages in thread
From: James E Wilson @ 2005-02-15 0:00 UTC (permalink / raw)
To: H. J. Lu; +Cc: jbeulich, binutils
On Sun, 2005-02-13 at 11:22, H. J. Lu wrote:
> * gas/ia64/ia64.exp: Add "operand-or".
> * gas/ia64/operand-or.d: New file.
> * gas/ia64/operand-or.s: Likewise.
Thanks for finding the problem and adding a testcase for it.
--
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2005-02-14 19:22 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-02-14 6:32 The new ia64 assembler is incompatible with glibc H. J. Lu
2005-02-14 8:35 ` H. J. Lu
2005-02-14 9:36 ` PATCH: Add an ia64 gas testcase for @nat|@qnan H. J. Lu
2005-02-15 0:00 ` James E Wilson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).