| Message ID | 20230221092726.420571-1-sam@gentoo.org |
|---|---|
| State | New |
| Headers | show |
| Series | stdio-common: tests: don't double-define _FORTIFY_SOURCE | expand |
On 2023-02-21 04:27, Sam James via Libc-alpha wrote: > Exactly the same as 35bcb08eaa953c9b8bef6ab2486dc4361e1f26c0. > > If using -D_FORITFY_SOURCE=3 (in my case, I've patched GCC to add > =3 instead of =2 (we've done =2 for years in Gentoo)), building > glibc tests will fail on tst-bz11319-fortify2 like: > ``` > <command-line>: error: "_FORTIFY_SOURCE" redefined [-Werror] > <built-in>: note: this is the location of the previous definition > cc1: all warnings being treated as errors > ``` > > It's just because we're always setting -D_FORTIFY_SOURCE=2 > rather than unsetting it first. If F_S is already 2, it's harmless, > but if it's another value (say, 1, or 3), the compiler will bawk. > > (I'm not aware of a reason this couldn't be tested with =3, > but the toolchain support is limited for that (too new), and we want > to run the tests everywhere possible.) > > As Siddhesh noted previously, we could implement some fallback > logic to determine the maximal F_S value supported by the toolchain, > which is a bit easier now that autoconf-archive has been updated for F_S=3 > (https://github.com/autoconf-archive/autoconf-archive/pull/269), but let's > revisit this if it continues to crop up. > > Signed-off-by: Sam James <sam@gentoo.org> > --- > stdio-common/Makefile | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org> Sam, if you're going to post patches regularly enough, please consider asking for write access. You could add me as a sponsor. https://sourceware.org/cgi-bin/pdw/ps_form.cgi Thanks, Sid > > diff --git a/stdio-common/Makefile b/stdio-common/Makefile > index 92a3499a94..abb398fdb7 100644 > --- a/stdio-common/Makefile > +++ b/stdio-common/Makefile > @@ -436,7 +436,7 @@ CFLAGS-tst-gets.c += -Wno-deprecated-declarations > > # BZ #11319 was first fixed for regular vdprintf, then reopened because > # the fortified version had the same bug. > -CFLAGS-tst-bz11319-fortify2.c += -D_FORTIFY_SOURCE=2 > +CFLAGS-tst-bz11319-fortify2.c += -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 > > CFLAGS-tst-memstream-string.c += -fno-builtin-fprintf >
Siddhesh Poyarekar <siddhesh@gotplt.org> writes: > On 2023-02-21 04:27, Sam James via Libc-alpha wrote: >> Exactly the same as 35bcb08eaa953c9b8bef6ab2486dc4361e1f26c0. >> If using -D_FORITFY_SOURCE=3 (in my case, I've patched GCC to add >> =3 instead of =2 (we've done =2 for years in Gentoo)), building >> glibc tests will fail on tst-bz11319-fortify2 like: >> ``` >> <command-line>: error: "_FORTIFY_SOURCE" redefined [-Werror] >> <built-in>: note: this is the location of the previous definition >> cc1: all warnings being treated as errors >> ``` >> It's just because we're always setting -D_FORTIFY_SOURCE=2 >> rather than unsetting it first. If F_S is already 2, it's harmless, >> but if it's another value (say, 1, or 3), the compiler will bawk. >> (I'm not aware of a reason this couldn't be tested with =3, >> but the toolchain support is limited for that (too new), and we want >> to run the tests everywhere possible.) >> As Siddhesh noted previously, we could implement some fallback >> logic to determine the maximal F_S value supported by the toolchain, >> which is a bit easier now that autoconf-archive has been updated for F_S=3 >> (https://github.com/autoconf-archive/autoconf-archive/pull/269), but let's >> revisit this if it continues to crop up. >> Signed-off-by: Sam James <sam@gentoo.org> >> --- >> stdio-common/Makefile | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) > > Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org> > > Sam, if you're going to post patches regularly enough, please consider > asking for write access. You could add me as a sponsor. > > https://sourceware.org/cgi-bin/pdw/ps_form.cgi Thanks Sid, very kind of you & much appreciated. All set up now! \o/ > > Thanks, > Sid > best, sam >> diff --git a/stdio-common/Makefile b/stdio-common/Makefile >> index 92a3499a94..abb398fdb7 100644 >> --- a/stdio-common/Makefile >> +++ b/stdio-common/Makefile >> @@ -436,7 +436,7 @@ CFLAGS-tst-gets.c += -Wno-deprecated-declarations >> # BZ #11319 was first fixed for regular vdprintf, then reopened >> because >> # the fortified version had the same bug. >> -CFLAGS-tst-bz11319-fortify2.c += -D_FORTIFY_SOURCE=2 >> +CFLAGS-tst-bz11319-fortify2.c += -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 >> CFLAGS-tst-memstream-string.c += -fno-builtin-fprintf >>
diff --git a/stdio-common/Makefile b/stdio-common/Makefile index 92a3499a94..abb398fdb7 100644 --- a/stdio-common/Makefile +++ b/stdio-common/Makefile @@ -436,7 +436,7 @@ CFLAGS-tst-gets.c += -Wno-deprecated-declarations # BZ #11319 was first fixed for regular vdprintf, then reopened because # the fortified version had the same bug. -CFLAGS-tst-bz11319-fortify2.c += -D_FORTIFY_SOURCE=2 +CFLAGS-tst-bz11319-fortify2.c += -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 CFLAGS-tst-memstream-string.c += -fno-builtin-fprintf
Exactly the same as 35bcb08eaa953c9b8bef6ab2486dc4361e1f26c0. If using -D_FORITFY_SOURCE=3 (in my case, I've patched GCC to add =3 instead of =2 (we've done =2 for years in Gentoo)), building glibc tests will fail on tst-bz11319-fortify2 like: ``` <command-line>: error: "_FORTIFY_SOURCE" redefined [-Werror] <built-in>: note: this is the location of the previous definition cc1: all warnings being treated as errors ``` It's just because we're always setting -D_FORTIFY_SOURCE=2 rather than unsetting it first. If F_S is already 2, it's harmless, but if it's another value (say, 1, or 3), the compiler will bawk. (I'm not aware of a reason this couldn't be tested with =3, but the toolchain support is limited for that (too new), and we want to run the tests everywhere possible.) As Siddhesh noted previously, we could implement some fallback logic to determine the maximal F_S value supported by the toolchain, which is a bit easier now that autoconf-archive has been updated for F_S=3 (https://github.com/autoconf-archive/autoconf-archive/pull/269), but let's revisit this if it continues to crop up. Signed-off-by: Sam James <sam@gentoo.org> --- stdio-common/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)