| Message ID | 20200306182419.13945-2-vgupta@synopsys.com |
|---|---|
| State | New |
| Headers | show |
| Series | glibc port to ARC processors | expand |
If this gets longlong.h back in sync with changes made in the GCC version, then please commit.
On 3/6/20 4:12 PM, Joseph Myers wrote: > If this gets longlong.h back in sync with changes made in the GCC version, > then please commit. Yes it does. The only differences now are in the initial licensing header due to glibc's variant bumping year and the http -> https. BTW, I don't think I asked for commit access before. I just rechecked MAINTAINERS wiki page and seems like someone needs to grant me that. Thx for taking a look. -Vineet
On 06/03/2020 21.24, Vineet Gupta wrote: > simple test such as below was failing. > > | void main(int argc, char *argv[]) > | { > | size_t total_time = 115424; // expected 115.424 > | double secs = (double)total_time/(double)1000; > | printf("%s %d %lf\n", "secs", total_time, secs); // prints 113.504 > | printf("%d\n", (size_t)secs); > | } > > The printf eventually called into glibc stdlib/divrem.c:__mpn_divrem() > which uses the __arc__ specific inline asm macros from longlong.h which > were causing miscompilation. > > include/ > 2019-03-28 Vineet Gupta <vgupta@synopsys.com> > > PR 89877 > > * longlong.h [__arc__] (add_ssaaaa): Add cc clobber > (sub_ddmmss): Likewise. Is PR number in subject (88409) wrong? The one in Changelog (89877) seems to be right.
On 3/9/20 8:23 AM, Alexander Cherepanov wrote: >> >> PR 89877 >> >> * longlong.h [__arc__] (add_ssaaaa): Add cc clobber >> (sub_ddmmss): Likewise. > > Is PR number in subject (88409) wrong? The one in Changelog (89877) seems to be > right. Yes you are right. I'll fix this up. Thx, -Vineet
On 3/6/20 4:21 PM, Vineet Gupta wrote: > > BTW, I don't think I asked for commit access before. I just rechecked MAINTAINERS > wiki page and seems like someone needs to grant me that. Hi Joseph, Carlos I'm trying to setup my sourceware account etc and the form below [1] needs "email address of person who approved request". Whats the typical procedure for getting that ? Thx, -Vineet [1] https://sourceware.org/cgi-bin/pdw/ps_form.cgi
On 3/10/20 1:23 PM, Vineet Gupta wrote: > On 3/6/20 4:21 PM, Vineet Gupta wrote: >> >> BTW, I don't think I asked for commit access before. I just rechecked MAINTAINERS >> wiki page and seems like someone needs to grant me that. > > Hi Joseph, Carlos > > I'm trying to setup my sourceware account etc and the form below [1] needs "email > address of person who approved request". Whats the typical procedure for getting > that ? Please take a look at the following page: "Becoming a maintainer (developer)": https://sourceware.org/glibc/wiki/MAINTAINERS#Becoming_a_maintainer_.28developer.29 ~~~ 5. Ask for commit access. * Ask one of the project stewards to authorize your commit access. * If the project steward authorizes you for commit access follow these instructions to create a sourceware account with commit access to the glibc repository. ~~~ Joseph and I are both stewards. "Project stewards (GNU package maintainers)" https://sourceware.org/glibc/wiki/MAINTAINERS#Project_stewards_.28GNU_package_maintainers.29 I would be happy to authorize your commit access. I assume your intent is to be one of the machine maintainers for ARC? > Thx, > -Vineet > > [1] https://sourceware.org/cgi-bin/pdw/ps_form.cgi >
On 3/10/20 10:57 AM, Carlos O'Donell via Libc-alpha wrote: > Please take a look at the following page: > > "Becoming a maintainer (developer)": > https://sourceware.org/glibc/wiki/MAINTAINERS#Becoming_a_maintainer_.28developer.29 > ~~~ > 5. Ask for commit access. > > * Ask one of the project stewards to authorize your commit access. > > * If the project steward authorizes you for commit access follow these > instructions to create a sourceware account with commit access to the > glibc repository. > ~~~ > > Joseph and I are both stewards. > "Project stewards (GNU package maintainers)" > https://sourceware.org/glibc/wiki/MAINTAINERS#Project_stewards_.28GNU_package_maintainers.29 > > I would be happy to authorize your commit access. > > I assume your intent is to be one of the machine maintainers for ARC? Yeah, this email was my indirect way of asking for commit access :-) Indeed this is for ARC glibc machine maintainer. Thx, -Vineet
On 3/10/20 2:08 PM, Vineet Gupta wrote: > On 3/10/20 10:57 AM, Carlos O'Donell via Libc-alpha wrote: >> Please take a look at the following page: >> >> "Becoming a maintainer (developer)": >> https://sourceware.org/glibc/wiki/MAINTAINERS#Becoming_a_maintainer_.28developer.29 >> ~~~ >> 5. Ask for commit access. >> >> * Ask one of the project stewards to authorize your commit access. >> >> * If the project steward authorizes you for commit access follow these >> instructions to create a sourceware account with commit access to the >> glibc repository. >> ~~~ >> >> Joseph and I are both stewards. >> "Project stewards (GNU package maintainers)" >> https://sourceware.org/glibc/wiki/MAINTAINERS#Project_stewards_.28GNU_package_maintainers.29 >> >> I would be happy to authorize your commit access. >> >> I assume your intent is to be one of the machine maintainers for ARC? > > Yeah, this email was my indirect way of asking for commit access :-) > Indeed this is for ARC glibc machine maintainer. Please submit the request then and include my email.
On 3/6/20 4:12 PM, Joseph Myers wrote: > If this gets longlong.h back in sync with changes made in the GCC version, > then please commit. This is now committed (my first commit yay !) Thx, -Vineet
diff --git a/stdlib/longlong.h b/stdlib/longlong.h index ee4aac1bb5a0..638b7894d48c 100644 --- a/stdlib/longlong.h +++ b/stdlib/longlong.h @@ -199,7 +199,8 @@ extern UDItype __udiv_qrnnd (UDItype *, UDItype, UDItype, UDItype); : "%r" ((USItype) (ah)), \ "rICal" ((USItype) (bh)), \ "%r" ((USItype) (al)), \ - "rICal" ((USItype) (bl))) + "rICal" ((USItype) (bl)) \ + : "cc") #define sub_ddmmss(sh, sl, ah, al, bh, bl) \ __asm__ ("sub.f %1, %4, %5\n\tsbc %0, %2, %3" \ : "=r" ((USItype) (sh)), \ @@ -207,7 +208,8 @@ extern UDItype __udiv_qrnnd (UDItype *, UDItype, UDItype, UDItype); : "r" ((USItype) (ah)), \ "rICal" ((USItype) (bh)), \ "r" ((USItype) (al)), \ - "rICal" ((USItype) (bl))) + "rICal" ((USItype) (bl)) \ + : "cc") #define __umulsidi3(u,v) ((UDItype)(USItype)u*(USItype)v) #ifdef __ARC_NORM__
simple test such as below was failing. | void main(int argc, char *argv[]) | { | size_t total_time = 115424; // expected 115.424 | double secs = (double)total_time/(double)1000; | printf("%s %d %lf\n", "secs", total_time, secs); // prints 113.504 | printf("%d\n", (size_t)secs); | } The printf eventually called into glibc stdlib/divrem.c:__mpn_divrem() which uses the __arc__ specific inline asm macros from longlong.h which were causing miscompilation. include/ 2019-03-28 Vineet Gupta <vgupta@synopsys.com> PR 89877 * longlong.h [__arc__] (add_ssaaaa): Add cc clobber (sub_ddmmss): Likewise. Signed-off-by: Vineet Gupta <vgupta@synopsys.com> --- stdlib/longlong.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)