GCC warnings
Albert Strasheim
With the build now using -Werror, I am having some build problems with
GCC 4.6 on Fedora 15.
tip currently produces the following warnings:
go/src/libmach/executable.c:994:11: warning: variable 'swab' set but
not used [-Wunused-but-set-variable]
go/src/cmd/6l/asm.c:684:29: warning: variable 'w' set but not used [-
Wunused-but-set-variable]
go/src/cmd/6l/asm.c:684:12: warning: variable 'va' set but not used [-
Wunused-but-set-variable]
../ld/data.c:879:38: warning: variable 'bss' set but not used [-
Wunused-but-set-variable]
../ld/dwarf.c:1382:16: warning: variable 'linksize' set but not used [-
Wunused-but-set-variable]
../ld/ldelf.c:322:13: warning: variable 'dp' set but not used [-
Wunused-but-set-variable]
go/src/cmd/6l/obj.c:383:21: warning: variable 'di' set but not used [-
Wunused-but-set-variable]
../ld/lib.c:1036:19: warning: variable 'last' set but not used [-
Wunused-but-set-variable]
go/src/cmd/gc/dcl.c:700:12: warning: variable 'init' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/reflect.c:140:6: warning: variable 'o' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/reflect.c:243:6: warning: variable 'o' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/swt.c:253:13: warning: variable 'oc' set but not used [-
Wunused-but-set-variable]
go/src/cmd/gc/swt.c:253:8: warning: variable 'os' set but not used [-
Wunused-but-set-variable]
go/src/cmd/gc/walk.c:200:12: warning: variable 'init' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/walk.c:383:10: warning: variable 'lno' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/subr.c:663:12: warning: variable 'init' set but not used
[-Wunused-but-set-variable]
go/src/cmd/gc/subr.c:1731:6: warning: variable 'ptr' set but not used
[-Wunused-but-set-variable]
go/src/cmd/6g/reg.c:1196:8: warning: variable 'p' set but not used [-
Wunused-but-set-variable]
This warning could be disabled, but I think it could turn up some
potential bugs.
Regards
Albert
Lorenzo Stoakes
on linux/386 I am getting a number of 'rm: cannot remove `build.out':
Text file busy' errors on build which seems to be preventing the
package from building correctly.
--
Lorenzo Stoakes
http://www.codegrunt.co.uk
peterGo
Is your problem related to this recent change?
build: tidy intermediate files during build
https://code.google.com/p/go/source/detail?r=3b7654f28a02511e4f9d3be392cab279ed8be37d
Peter
On Apr 14, 6:00 am, Lorenzo Stoakes <lstoa...@gmail.com> wrote:
> This may be something to do with my particular configuration, however
> on linux/386 I am getting a number of 'rm: cannot remove `build.out':
> Text file busy' errors on build which seems to be preventing the
> package from building correctly.
>
Lorenzo Stoakes
> Lorenzo,
>
> Is your problem related to this recent change?
>
> build: tidy intermediate files during build
> https://code.google.com/p/go/source/detail?r=3b7654f28a02511e4f9d3be392cab279ed8be37d
>
Potentially; I am briefly using a mac where this problem does *not*
arise, once I am back on the linux box I shall investigate :) - I'd
certainly be surprised if it was the -Werror change, as that seems
unrelated.
Cheers, Lorenzo
peterGo
I've just successfully run ./all.bash on linux/386 (Ubuntu 10.10) hg
id 165b8254aee1.
Peter
On Apr 14, 8:09 am, Lorenzo Stoakes <lstoa...@gmail.com> wrote:
>
> > Lorenzo,
>
> > Is your problem related to this recent change?
>
> > build: tidy intermediate files during build
Adam Langley
> Potentially; I am briefly using a mac where this problem does *not*
> arise, once I am back on the linux box I shall investigate :) - I'd
> certainly be surprised if it was the -Werror change, as that seems
> unrelated.
Is NFS involved when the problem occurs?
AGL
Lorenzo Stoakes
> Is NFS involved when the problem occurs?
No, 'vboxsf' - I am running linux in a virtualbox vm and the go folder
is sat in a shared folder from the host system, which itself is sat in
a truecrypt volume, so it's pretty sloooow.
uname -a gives:-
Linux loki 2.6.35-28-generic #49-Ubuntu SMP Tue Mar 1 14:40:58 UTC 2011 i686 GNU
/Linux
I will try the revision before the CL Peter mentions. 165b8254aee1
does *not* build for me.
Lorenzo Stoakes
builds but 3b7654f28a02 does not.
Russ Cox
> This may be something to do with my particular configuration, however
> on linux/386 I am getting a number of 'rm: cannot remove `build.out':
> Text file busy' errors on build which seems to be preventing the
> package from building correctly.
What do you mean by "seems to"? Does the build stop or not?
It does appear that the build is now removing build.out
while it is still open for reading, but that is typically allowed
by Unix file systems. Even if it produced an error, the
command that is doing the remove is rm -f, which should
still exit successfully in that case.
Russ
Lorenzo Stoakes
> What do you mean by "seems to"? Does the build stop or not?
Sorry I meant to say that that seems to be the *cause* of the build
failing - yes the build does stop.
>
> It does appear that the build is now removing build.out
> while it is still open for reading, but that is typically allowed
> by Unix file systems. Even if it produced an error, the
> command that is doing the remove is rm -f, which should
> still exit successfully in that case.
This could be a peculiarity of vboxsf (virtual box shared folder), I
even tried alias rm='rm -f' with no effect (before I knew the build
used -f anyway).I will try copying the folder to a non-shared folder
and building.
Lorenzo Stoakes
> This could be a peculiarity of vboxsf (virtual box shared folder), I
> even tried alias rm='rm -f' with no effect (before I knew the build
> used -f anyway).I will try copying the folder to a non-shared folder
> and building.
This worked; clearly related to vboxsf. Very strange; this seems to be
a known issue with vboxsf, however -
http://askubuntu.com/questions/22824/cant-modify-files-in-my-shared-folder-virtualbox-guest-is-ubuntu-host-is-wind
(I am using a windows host on this box)
Russ Cox
from the build failure.
Russ
Lorenzo Stoakes
%%%% making pkg %%%%
make: Entering directory `/home/lorenzo/Dropbox/code/go/src/pkg'
install runtime
install sync/atomic
install unicode
install math
INSTALL FAIL sync/atomic
make[1]: Entering directory `/home/lorenzo/Dropbox/code/go/src/pkg/sync/atomic'
8g -o _go_.8 doc.go
8a asm_386.s
rm -f _obj/sync/atomic.a
gopack grc _obj/sync/atomic.a _go_.8 asm_386.8
cp _obj/sync/atomic.a
"/home/lorenzo/Dropbox/code/go/pkg/linux_386/sync/atomic.a"
rm -rf *.o *.a *.[568vq] [568vq].out *.so _obj _test _testmain.go
*.exe _cgo* *.cgo[12].* test.out build.out
rm: cannot remove `build.out': Text file busy
make[1]: *** [install.clean] Error 1
make[1]: Leaving directory `/home/lorenzo/Dropbox/code/go/src/pkg/sync/atomic'
make: *** [sync/atomic.install] Error 1
make: *** Waiting for unfinished jobs....
INSTALL FAIL unicode
make[1]: Entering directory `/home/lorenzo/Dropbox/code/go/src/pkg/unicode'
8g -o _go_.8 casetables.go digit.go letter.go tables.go
rm -f _obj/unicode.a
gopack grc _obj/unicode.a _go_.8
cp _obj/unicode.a "/home/lorenzo/Dropbox/code/go/pkg/linux_386/unicode.a"
rm -rf *.o *.a *.[568vq] [568vq].out *.so _obj _test _testmain.go
*.exe _cgo* *.cgo[12].* test.out build.out maketables
rm: cannot remove `build.out': Text file busy
make[1]: *** [install.clean] Error 1
make[1]: Leaving directory `/home/lorenzo/Dropbox/code/go/src/pkg/unicode'
make: *** [unicode.install] Error 1
INSTALL FAIL math
make[1]: Entering directory `/home/lorenzo/Dropbox/code/go/src/pkg/math'
8g -o _go_.8 acosh.go asinh.go atanh.go bits.go cbrt.go const.go
copysign.go erf.go exp_port.go fdim.go gamma.go hypot_port.go j0.go
j1.go jn.go lgamma.go logb.go nextafter.go pow.go pow10.go signbit.go
sinh.go sqrt_port.go tanh.go unsafe.go asin_decl.go atan_decl.go
atan2_decl.go exp_decl.go exp2_decl.go expm1_decl.go fabs_decl.go
floor_decl.go frexp_decl.go fmod_decl.go hypot_decl.go ldexp_decl.go
log_decl.go log10_decl.go log1p_decl.go modf_decl.go remainder_decl.go
sin_decl.go sincos_decl.go sqrt_decl.go tan_decl.go
8a asin_386.s
8a atan_386.s
8a atan2_386.s
8a exp_386.s
8a exp2_386.s
8a expm1_386.s
8a fabs_386.s
8a floor_386.s
8a frexp_386.s
8a fmod_386.s
8a hypot_386.s
8a ldexp_386.s
8a log_386.s
8a log10_386.s
8a log1p_386.s
8a modf_386.s
8a remainder_386.s
8a sin_386.s
8a sincos_386.s
8a sqrt_386.s
8a tan_386.s
rm -f _obj/math.a
gopack grc _obj/math.a _go_.8 asin_386.8 atan_386.8 atan2_386.8
exp_386.8 exp2_386.8 expm1_386.8 fabs_386.8 floor_386.8 frexp_386.8
fmod_386.8 hypot_386.8 ldexp_386.8 log_386.8 log10_386.8 log1p_386.8
modf_386.8 remainder_386.8 sin_386.8 sincos_386.8 sqrt_386.8 tan_386.8
cp _obj/math.a "/home/lorenzo/Dropbox/code/go/pkg/linux_386/math.a"
rm -rf *.o *.a *.[568vq] [568vq].out *.so _obj _test _testmain.go
*.exe _cgo* *.cgo[12].* test.out build.out
rm: cannot remove `build.out': Text file busy
make[1]: *** [install.clean] Error 1
make[1]: Leaving directory `/home/lorenzo/Dropbox/code/go/src/pkg/math'
make: *** [math.install] Error 1
INSTALL FAIL runtime
make[1]: Entering directory `/home/lorenzo/Dropbox/code/go/src/pkg/runtime'
cp runtime.h "/home/lorenzo/Dropbox/code/go/pkg/linux_386/runtime.h"
cp cgocall.h "/home/lorenzo/Dropbox/code/go/pkg/linux_386/cgocall.h"
quietgcc -o mkversion -I "/home/lorenzo/Dropbox/code/go/include"
mkversion.c "/home/lorenzo/Dropbox/code/go/lib/lib9.a"
(echo 'package runtime'; echo 'const theGoos = "linux"') >version_linux.go
(echo 'package runtime'; echo 'const theGoarch = "386"') >version_386.go
CC="8c" CFLAGS="-Ilinux -I386 -Ilinux/386 -wF " ./mkgodefs.sh proc.c
iface.c hashmap.c chan.c > runtime_defs.go.x
mv -f runtime_defs.go.x runtime_defs.go
8c -Ilinux -I386 -Ilinux/386 -wF -a proc.c >runtime.acid.386
quietgcc -o goc2c goc2c.c
8c -Ilinux -I386 -Ilinux/386 -wF amd64/traceback.c
GOROOT="/home/lorenzo/Dropbox/code/go" ./mkversion >version.go
./mkasmh.sh >386/asm.h.x
mv -f 386/asm.h.x 386/asm.h
8g -+ -o _go_.8 debug.go error.go extern.go mem.go sig.go
softfloat64.go type.go version.go version_linux.go version_386.go
runtime_defs.go
8a 386/asm.s
8c -Ilinux -I386 -Ilinux/386 -wF -o "cgocall.8" cgocall.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "chan.8" chan.c
8c -Ilinux -I386 -Ilinux/386 -wF 386/closure.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "cpuprof.8" cpuprof.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "float.8" float.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "complex.8" complex.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "hashmap.8" hashmap.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "iface.8" iface.c
./goc2c "`pwd`/malloc.goc" > malloc.c.tmp
8c -Ilinux -I386 -Ilinux/386 -wF -o "mcache.8" mcache.c
mv -f malloc.c.tmp malloc.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "mcentral.8" mcentral.c
8c -Ilinux -I386 -Ilinux/386 -wF linux/mem.c
8a 386/memmove.s
8c -Ilinux -I386 -Ilinux/386 -wF -o "mfinal.8" mfinal.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "mfixalloc.8" mfixalloc.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "mgc0.8" mgc0.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "mheap.8" mheap.c
./goc2c "`pwd`/mprof.goc" > mprof.c.tmp
8c -Ilinux -I386 -Ilinux/386 -wF -o "msize.8" msize.c
mv -f mprof.c.tmp mprof.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "print.8" print.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "proc.8" proc.c
./goc2c "`pwd`/reflect.goc" > reflect.c.tmp
8c -Ilinux -I386 -Ilinux/386 -wF -o "rune.8" rune.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "runtime.8" runtime.c
mv -f reflect.c.tmp reflect.c
./goc2c "`pwd`/runtime1.goc" > runtime1.c.tmp
8a linux/386/rt0.s
./goc2c "`pwd`/sema.goc" > sema.c.tmp
8c -Ilinux -I386 -Ilinux/386 -wF linux/386/signal.c
mv -f runtime1.c.tmp runtime1.c
./goc2c "`pwd`/sigqueue.goc" > sigqueue.c.tmp
mv -f sema.c.tmp sema.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "slice.8" slice.c
./goc2c "`pwd`/string.goc" > string.c.tmp
8c -Ilinux -I386 -Ilinux/386 -wF -o "symtab.8" symtab.c
mv -f sigqueue.c.tmp sigqueue.c
8a linux/386/sys.s
mv -f string.c.tmp string.c
8c -Ilinux -I386 -Ilinux/386 -wF linux/thread.c
8a 386/vlop.s
8c -Ilinux -I386 -Ilinux/386 -wF 386/vlrt.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "malloc.8" malloc.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "mprof.8" mprof.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "reflect.8" reflect.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "runtime1.8" runtime1.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "sema.8" sema.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "sigqueue.8" sigqueue.c
8c -Ilinux -I386 -Ilinux/386 -wF -o "string.8" string.c
rm -f _obj/runtime.a
gopack grc _obj/runtime.a _go_.8 asm.8 cgocall.8 chan.8 closure.8
cpuprof.8 float.8 complex.8 hashmap.8 iface.8 malloc.8 mcache.8
mcentral.8 mem.8 memmove.8 mfinal.8 mfixalloc.8 mgc0.8 mheap.8 mprof.8
msize.8 print.8 proc.8 reflect.8 rune.8 runtime.8 runtime1.8 rt0.8
sema.8 signal.8 sigqueue.8 slice.8 string.8 symtab.8 sys.8 thread.8
traceback.8 vlop.8 vlrt.8
cp _obj/runtime.a "/home/lorenzo/Dropbox/code/go/pkg/linux_386/runtime.a"
rm -rf *.o *.a *.[568vq] [568vq].out version.go version_*.go *.so _obj
_test _testmain.go *.exe _cgo* *.cgo[12].* test.out build.out
rm: cannot remove `build.out': Text file busy
make[1]: *** [install.clean] Error 1
rm reflect.c string.c sema.c sigqueue.c runtime1.c malloc.c mprof.c
make[1]: Leaving directory `/home/lorenzo/Dropbox/code/go/src/pkg/runtime'
make: *** [runtime.install] Error 1
make: Leaving directory `/home/lorenzo/Dropbox/code/go/src/pkg'
Russ Cox
what happens if you run
rm -rf build.out >build.out
echo $?
Dave Cheney
Are you able to demonstrate the build failure outside a path that is monitored by some sort of external file sync tool?
Cheers
Dave
Sent from my C64
Lorenzo Stoakes
lorenzo@loki:/home/lorenzo/Dropbox/code/go/src/cmd$ rm -rf build.out>build.out
rm: cannot remove `build.out': Text file busy
lorenzo@loki:/home/lorenzo/Dropbox/code/go/src/cmd$ echo $?
1
Lorenzo Stoakes
> Hi Lorenzo
>
> Are you able to demonstrate the build failure outside a path that is monitored by some sort of external file sync tool?
>
I will try this and get back to you.
Lorenzo Stoakes
> On 14 April 2011 15:38, Dave Cheney <da...@cheney.net> wrote:
>> Hi Lorenzo
>>
>> Are you able to demonstrate the build failure outside a path that is monitored by some sort of external file sync tool?
>>
>
> I will try this and get back to you.
Tried it out - still fails. Looks like it's definitely vboxsf.
Lorenzo Stoakes
line; since this is simply a quick optimisation, this prevents the
build breaking in *weird* situations like my vboxsf folder.
I've submitted a trivial CL for this at http://codereview.appspot.com/4412045
Cheers, Lorenzo
Ian Lance Taylor
> It does appear that the build is now removing build.out
> while it is still open for reading, but that is typically allowed
> by Unix file systems. Even if it produced an error, the
> command that is doing the remove is rm -f, which should
> still exit successfully in that case.
I believe POSIX says that rm -f must exit successfully if the file does
not exist, but that if the file exists but removing it fails, rm is
supposed to exit with a failure status. That is, -f means ignore ENOENT
but does not mean ignore other errors.
Ian
Lorenzo Stoakes
> I believe POSIX says that rm -f must exit successfully if the file does
> not exist, but that if the file exists but removing it fails, rm is
> supposed to exit with a failure status. That is, -f means ignore ENOENT
> but does not mean ignore other errors.
I believe the error here is ETXTBSY, hence it not being ignored.
r
go through this list and clean them up.
Thanks for the detailed list.
-rob
Albert Strasheim
On Apr 14, 6:22 pm, r <r...@google.com> wrote:
> Disable the flag for now if it's giving you trouble. Later today I'll
> go through this list and clean them up.
> Thanks for the detailed list.
list. Might be that GCC only warns once per line.
I think this one might also be fixed by another CL that hasn't been
checked in yet.
Anyway:
go/src/cmd/6l/asm.c:684:21: warning: variable 'fo' set but not used [-
Wunused-but-set-variable]
Regards
Albert