This is the mail archive of the libc-ports@sources.redhat.com mailing list for the libc-ports project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH 1/2] AArch64 glibc port


Folks,

This revised patch addresses the comments raised by Carlos, Andreas
and Joseph on the original aarch64 port submission.

The main changes since from the original patch are:

* Copyright years have been merged inline with the documentation
Carlos highlighted.
* ARCH_PLTEXIT_MEMBERS prototype corrected.
* pthread_spin_lock and pthread_spin_trylock switched to generic versions.
* The port has been rebased over the tree as of 1/11/2012, this brings
in a number of the changes highlighted as missing by Joseph.
* ADDRIDX, VALIDX  and VERSYMIDX definitions removed.
* Added sotruss-lib.c
* Added get-rounding-mode.h
* Refactor around recent clean up of init-first.c
* Refactor around recent cleanup of fcntl.h
* libm-test-ulps regenerated, bringing in clog definitions and setting
sqrt ulps to 0.
* Relocated FPU support to fpu/ directory.
* Remove CFLAGS-tst-backtrace* cruft.
* Remove CHECK_SP cruft.
* FE_INVALID and friends defined using the same mechanism adopted in
other ports.
* fenv_t members renamed to avoid namespace pollution
* FP_FAST_FMA and FP_FAST_FMAF hardwired on.
* FP_TRAPPING_EXCEPTIONS defined
* Default version moved to GLIBC_2.17
* Minimum kernel version moved to 3.7.0
* Assorted port specific macros removed.
* Switch to generic not-cancel.h
* Switch to generic syscalls.list
* Removed #undef NR_* cruft.
* Duplicate Makefile entry for epoll_create and friend removed.

The -fexception CFLAGS change has been removed and posted to
libc-alpha independently.

Joseph's review pointed out that the port should switch from __sync to
__atomic primitives, work on this is in progress, but not included in
this patch.

The test suite has been run cross, we have the following outstanding
issues that need further investigation:

* test-ldouble.out
  27 instances of underflow set.  These all appear to be related to
scalbn setting underflow when not expected.

* test-double.out
  38 instances of underflow not set on fma related tests.

* test-ildoubl.out
  3 errors for ULPs on csqrt tests.

Updated ChangeLog attached.

Thanks
/Marcus

Attachment: glibc-4078da-v2-changelog.txt
Description: Text document

Attachment: glibc-4078da-v2.diff
Description: Binary data


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]