Re:architectures, please report on trunk status

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re:architectures, please report on trunk status

Steven Munroe
Roland McGrath <[hidden email]> wrote on 02/28/2006 12:21:00 PM:

> People responsible for the architectures in the main source tree, and
> people who build and test on them, please report ASAP on their status.

powerpc64 builds and runs make check including -mlong-double-128 and nptl.
powerpc32 builds and fails make check in "run_iconv_tests.sh" with a segsegv, investigating.

> verify that "make check-abi" results show no old symbols gone missing.
> Hurd folks, berate me privately.

for powerpc64 make check-abi reports:

LC_ALL=C gawk -f ../scripts/extract-abilist.awk -v 'config=powerpc64-unknown-linux-gnu/thread'  ../abilist/libm.abilist | { diff -p -U 0 - /home/sjmunroe/work/build-24x/math/libm.symlist  ; }
--- - 2006-02-28 17:45:08.361584000 -0600
+++ /home/sjmunroe/work/build-24x/math/libm.symlist 2006-02-28 17:45:08.349404984 -0600
@@ -317,0 +318,104 @@ GLIBC_2.3
+GLIBC_2.4
+ GLIBC_2.4 A
+ __clog10l F
+ __finitel F
+ __fpclassifyl F
+ __nldbl_nexttowardf F
+ __signbitl F
+ acoshl F
+ acosl F
+ asinhl F
+ asinl F
+ atan2l F
+ atanhl F
+ atanl F
+ cabsl F
+ cacoshl F
+ cacosl F
+ cargl F
+ casinhl F
+ casinl F
+ catanhl F
+ catanl F
+ cbrtl F
+ ccoshl F
+ ccosl F
+ ceill F
+ cexpl F
+ cimagl F
+ clog10l F
+ clogl F
+ conjl F
+ copysignl F
+ coshl F
+ cosl F
+ cpowl F
+ cprojl F
+ creall F
+ csinhl F
+ csinl F
+ csqrtl F
+ ctanhl F
+ ctanl F
+ dreml F
+ erfcl F
+ erfl F
+ exp10l F
+ exp2l F
+ expl F
+ expm1l F
+ fabsl F
+ fdiml F
+ finitel F
+ floorl F
+ fmal F
+ fmaxl F
+ fminl F
+ fmodl F
+ frexpl F
+ gammal F
+ hypotl F
+ ilogbl F
+ j0l F
+ j1l F
+ jnl F
+ ldexpl F
+ lgammal F
+ lgammal_r F
+ llrintl F
+ llroundl F
+ log10l F
+ log1pl F
+ log2l F
+ logbl F
+ logl F
+ lrintl F
+ lroundl F
+ modfl F
+ nanl F
+ nearbyintl F
+ nextafterl F
+ nexttoward F
+ nexttowardf F
+ nexttowardl F
+ pow10l F
+ powl F
+ remainderl F
+ remquol F
+ rintl F
+ roundl F
+ scalbl F
+ scalblnl F
+ scalbnl F
+ significandl F
+ sincosl F
+ sinhl F
+ sinl F
+ sqrtl F
+ tanhl F
+ tanl F
+ tgammal F
+ truncl F
+ y0l F
+ y1l F
+ ynl F
make[2]: *** [check-abi-libm] Error 1
rm /home/sjmunroe/work/build-24x/math/libm.dynsym /home/sjmunroe/work/build-24x/math/libm.symlist

I assume this is just an artifact of the long-double abi change (adding symbols, no removes). So this is OK?

These runs where from a POWER5 p570, running a 2.6.5 kernel and gcc-4.1. More tests on different hardware and kernels comming...