Mercurialcrosstool-ng / file revision
summary | shortlog | changelog | graph | tags | branches | files | changeset | file | latest | revisions | annotate | diff | raw | help
patches/glibc/ports-2.10.1/120-i386-x86_64-revert-clone-cfi.patch
author Anthony Foiani <anthony.foiani@gmail.com>
Thu May 19 23:06:16 2011 +0200 (2011-05-19)
changeset 2461 ec30b191f0e3
permissions -rw-r--r--
complibs/ppl: build only C and C++ interfaces for PPL

By default, PPL wants to build interfaces for any of a variety of
langauges it finds on the local host (python, java, possibly perl, also
more esoteric languages such as ocaml and prolog).

These extra interfaces can double the compile time for the library. For
single-process builds, I found a savings of more than 40%:

default / j1: 716s total, 143.2s avg, 0.52s stdev
just_c / j1: 406s total, 81.2s avg, 0.33s stdev
just_c_cpp / j1: 413s total, 82.6s avg, 0.22s stdev

And for multi-process builds, it approached 50%:

default / j4: 625s total, 125.0s avg, 0.57s stdev
just_c / j4: 338s total, 67.6s avg, 1.25s stdev
just_c_cpp / j4: 327s total, 65.4s avg, 0.36s stdev

Since the PPL we build within ct-ng is only used by GCC, we only need to
build the C and C++ interfaces.

Signed-Off-By: Anthony Foiani <anthony.foiani@gmail.com> 
     1 revert cfi additions to clone on i386/x86_64 to workaround problems in

     2 gcc's unwinder code.  this is not a bug in glibc, it triggers problems

     3 elsewhere.  this cfi code does not gain us a whole lot anyways.

     4 

     5 http://gcc.gnu.org/ml/gcc/2006-12/msg00293.html

     6 

     7 Index: sysdeps/unix/sysv/linux/i386/clone.S

     8 ===================================================================

     9 RCS file: /cvs/glibc/libc/sysdeps/unix/sysv/linux/i386/clone.S,v

    10 retrieving revision 1.27

    11 retrieving revision 1.26

    12 diff -u -p -r1.27 -r1.26

    13 

    14 diff -durN glibc-2.10.1.orig/sysdeps/unix/sysv/linux/i386/clone.S glibc-2.10.1/sysdeps/unix/sysv/linux/i386/clone.S

    15 --- glibc-2.10.1.orig/sysdeps/unix/sysv/linux/i386/clone.S	2006-12-04 00:12:36.000000000 +0100

    16 +++ glibc-2.10.1/sysdeps/unix/sysv/linux/i386/clone.S	2009-11-13 00:49:45.000000000 +0100

    17 @@ -120,9 +120,6 @@

    18  	ret

    19  

    20  L(thread_start):

    21 -	cfi_startproc;

    22 -	/* Clearing frame pointer is insufficient, use CFI.  */

    23 -	cfi_undefined (eip);

    24  	/* Note: %esi is zero.  */

    25  	movl	%esi,%ebp	/* terminate the stack frame */

    26  #ifdef RESET_PID

    27 @@ -155,7 +152,6 @@

    28  	jmp	L(haspid)

    29  	.previous

    30  #endif

    31 -	cfi_endproc;

    32  

    33  	cfi_startproc

    34  PSEUDO_END (BP_SYM (__clone))

    35 diff -durN glibc-2.10.1.orig/sysdeps/unix/sysv/linux/x86_64/clone.S glibc-2.10.1/sysdeps/unix/sysv/linux/x86_64/clone.S

    36 --- glibc-2.10.1.orig/sysdeps/unix/sysv/linux/x86_64/clone.S	2006-12-04 00:12:36.000000000 +0100

    37 +++ glibc-2.10.1/sysdeps/unix/sysv/linux/x86_64/clone.S	2009-11-13 00:49:45.000000000 +0100

    38 @@ -89,9 +89,6 @@

    39  	ret

    40  

    41  L(thread_start):

    42 -	cfi_startproc;

    43 -	/* Clearing frame pointer is insufficient, use CFI.  */

    44 -	cfi_undefined (rip);

    45  	/* Clear the frame pointer.  The ABI suggests this be done, to mark

    46  	   the outermost frame obviously.  */

    47  	xorl	%ebp, %ebp

    48 @@ -116,7 +113,6 @@

    49  	/* Call exit with return value from function call. */

    50  	movq	%rax, %rdi

    51  	call	HIDDEN_JUMPTARGET (_exit)

    52 -	cfi_endproc;

    53  

    54  	cfi_startproc;

    55  PSEUDO_END (BP_SYM (__clone))
crosstool-ng