[arch-dev-public] SBCL, kernel, and GCC 4.30
While trying to build the latest SBCL I came across an interesting problem. The build halts midway and clocks the CPU at 100%. A quick search shows that GCC 4.3.0 changed the ABI slightly to conform to the formal standard as opposed to the defacto standard. GCC stopped clearing or setting the DF before function calls which is out of step with the kernel which expects it set/cleared. Apparently this mostly affects signal handlers called by the kernel. This causes problems with the signal functions in our glibc because it was compiled with GCC 4.3.0. I installed glibc 2.7-7 and sbcl builds fine now. A better explanation of the problem can be found at http://lwn.net/Articles/272048/ and at the very bottem there is a link to a Debian bug. The problem is fixed in kernel 2.6.25 and (sigh) there is a patch available for 2.6.24. k -- K. Piche <kpiche@rogers.com>
On Sat, Mar 29, 2008 at 1:50 PM, K. Piche <kpiche@rogers.com> wrote:
While trying to build the latest SBCL I came across an interesting problem. The build halts midway and clocks the CPU at 100%.
A quick search shows that GCC 4.3.0 changed the ABI slightly to conform to the formal standard as opposed to the defacto standard. GCC stopped clearing or setting the DF before function calls which is out of step with the kernel which expects it set/cleared. Apparently this mostly affects signal handlers called by the kernel. This causes problems with the signal functions in our glibc because it was compiled with GCC 4.3.0. I installed glibc 2.7-7 and sbcl builds fine now.
A better explanation of the problem can be found at http://lwn.net/Articles/272048/ and at the very bottem there is a link to a Debian bug.
The problem is fixed in kernel 2.6.25 and (sigh) there is a patch available for 2.6.24.
This was fixed in 2.6.24.4 as well, so we shouldn't be seeing issues... http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.24.y.git;a=commit... -Dan
Am Samstag, 29. März 2008 19:56:30 schrieb Dan McGee:
This was fixed in 2.6.24.4 as well, so we shouldn't be seeing issues...
Afaik the kernel in [testing] should not be affected. -- archlinux.de
On Sat, Mar 29, 2008 at 2:03 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Samstag, 29. März 2008 19:56:30 schrieb Dan McGee:
This was fixed in 2.6.24.4 as well, so we shouldn't be seeing issues...
Afaik the kernel in [testing] should not be affected.
Ahh, I didn't realize no 2.6.24.4 kernel ever made it out of testing- my mistake. Kevin, if you upgrade to the newest testing kernel you should not come across these errors. -Dan
Alright. I was doing the build from my core (non-testing) environment. k On Sat, 2008-03-29 at 14:17 -0500, Dan McGee wrote:
On Sat, Mar 29, 2008 at 2:03 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Samstag, 29. März 2008 19:56:30 schrieb Dan McGee:
This was fixed in 2.6.24.4 as well, so we shouldn't be seeing issues...
Afaik the kernel in [testing] should not be affected.
Ahh, I didn't realize no 2.6.24.4 kernel ever made it out of testing- my mistake. Kevin, if you upgrade to the newest testing kernel you should not come across these errors.
-Dan -- K. Piche <kpiche@rogers.com>
participants (3)
-
Dan McGee
-
K. Piche
-
Pierre Schmitz