See patch.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Uwe.
Uwe Hermann wrote:
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less often.
Signed-off-by: Uwe Hermann uwe@hermann-uwe.de
Acked-by: Peter Stuge peter@stuge.se
Hi Eric,
are there any reasons (bugs) to avoid using -O2 for ROMCC?
On 10.12.2008 11:19, Uwe Hermann wrote:
See patch.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less often.
Signed-off-by: Uwe Hermann uwe@hermann-uwe.de
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net pending an answer from Eric.
Regards, Carl-Daniel
Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net writes:
Hi Eric,
are there any reasons (bugs) to avoid using -O2 for ROMCC?
On 10.12.2008 11:19, Uwe Hermann wrote:
See patch.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less often.
Signed-off-by: Uwe Hermann uwe@hermann-uwe.de
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net pending an answer from Eric.
I want to say to my knowledge all of the failures in romcc kill the build. I can remember one very rare case where that was not true.
-mcpu=p2 should always be safe, and I would not expect any problems with -O2.
Eric
On 10.12.2008 14:32, Eric W. Biederman wrote:
Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net writes:
Hi Eric,
are there any reasons (bugs) to avoid using -O2 for ROMCC?
On 10.12.2008 11:19, Uwe Hermann wrote:
See patch.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less often.
Signed-off-by: Uwe Hermann uwe@hermann-uwe.de
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net pending an answer from Eric.
I want to say to my knowledge all of the failures in romcc kill the build. I can remember one very rare case where that was not true.
-mcpu=p2 should always be safe, and I would not expect any problems with -O2.
Thanks for confirming this!
Regards, Carl-Daniel
On Wed, Dec 10, 2008 at 03:41:10PM +0100, Carl-Daniel Hailfinger wrote:
On 10.12.2008 14:32, Eric W. Biederman wrote:
Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net writes:
Hi Eric,
are there any reasons (bugs) to avoid using -O2 for ROMCC?
On 10.12.2008 11:19, Uwe Hermann wrote:
See patch.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less often.
Signed-off-by: Uwe Hermann uwe@hermann-uwe.de
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net pending an answer from Eric.
I want to say to my knowledge all of the failures in romcc kill the build. I can remember one very rare case where that was not true.
-mcpu=p2 should always be safe, and I would not expect any problems with -O2.
Thanks for confirming this!
Yep, thanks a lot! Committed in r3810.
Uwe.
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less
often.
Hmm. So the -02 enables optimizations in romcc? Should we worry about this in -mcpu=p3 boards?
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is pending an answer from Eric.
I didn't think the Intel chips could do CAR, or I thought it hadn't been figured out yet?
On 10.12.2008 18:41, Joseph Smith wrote:
Build-tested and tested on hardware using an ASUS P2B-F, I didn't notice any problems. However, if there are any non-obvious reasons to _not_ enable optimizations in romcc, please let me know.
Use -O2 and -mcpu=p2 as romcc options for all Intel 440BX boards.
This should hopefully make the "too few registers" error pop up less
often.
Hmm. So the -02 enables optimizations in romcc? Should we worry about this in -mcpu=p3 boards?
I'd prefer to switch all these boards to CAR, but as this patch improves the current situation, it is pending an answer from Eric.
I didn't think the Intel chips could do CAR, or I thought it hadn't been figured out yet?
The original CAR code was written by an Intel intern IIRC and it was for old Intel CPUs. It is interesting to know that our current i586 CAR has almost nothing in common with the i586 CAR which was committed in r1967.
Regards, Carl-Daniel
-
Carl-Daniel Hailfinger -
ebiederm＠xmission.com -
Joseph Smith -
Peter Stuge -
Uwe Hermann