[U-Boot-Users] Patches for ARM IntegratorCP, ARM926EJS, and ARM946ES
Hi,
The patches were developed on an ARM IntegratorCP board with ARM946ES. Since the majority code of ARM946ES cpu was copied from ARM926EJS and there was bug about CFG_HZ found in ARM946ES later, I also fixed this CFG_HZ error on ARM926EJS.
The major changes are 1. Mulitple core modules support (u-boot.multicm.patch) Following OMAP boards' methodology, I added mulitple core module configuration. For example, integratorcp_946es_config is for IntegratorCP board with ARM946ES, integratorcp_config is for IntegratorCP board with ARM926EJS by default, integratorap_946es_config is for IntegratorAP board with ARM946ES, and integratorap_config is for IntegratorAP with ARM926EJS. Now only two core modules are supported, ARM946ES and ARM926EJS.
2. ARM946ES support (u-boot.arm946es.patch)
3. ARM926EJS bug fix (u-boot.arm926ejs.patch) Fixed CFG_HZ problem. Set CFG_HZ to 1000. The solution is from AT91RM9200.
4. IntegratorCP bug fixes (u-boot.integratorcp.patch) Added SDRAM size information and FLASH support.
Philippe, I don't have an ARM926EJS core module. Would you please test the CFG_HZ fix on your system? Thanks.
Regards, -Shawn.
Wolfgang,
Forgot to mention that the patches are againt the latest CVS code I checked out this morning.
Regards, -Shawn.
Wolfgang, Philippe, Peter and others,
I made some mistakes when dividing the whole patch into several logical sets of individual patches. The u-boot.arm926ejs.patch was totally wrong. I apologize here for my carelessness.
The patches attached today is againt the latest CVS (1.1.3) and I've tested to make sure there are no more such errors. Wolfgang, please be aware of the errors in the previous patches submitted a month ago (Dec. 20, 2004) and delete it from your backlog if possible.
Philippe and Peter, sorry for the trouble it might cause and waste your time. If having time, you can validate the patches on your platform, especially the patch for ARM926EJS.
Peter, thanks for the patch for the wrong configured timer value. I cannot understand why I set it to 0x8C instead of 0x8A at that moment. However please be aware that that patch was kinda obsolete. The better way to fix CFG_HZ issue (thus timer issue) is implemented in the patches attached today, which set CFG_HZ to 1000.
Best regards, -Shawn.
On Mon, 20 Dec 2004 10:55:06 -0800, Shawn Jin shawnxjin@gmail.com wrote:
Hi,
The patches were developed on an ARM IntegratorCP board with ARM946ES. Since the majority code of ARM946ES cpu was copied from ARM926EJS and there was bug about CFG_HZ found in ARM946ES later, I also fixed this CFG_HZ error on ARM926EJS.
The major changes are
- Mulitple core modules support (u-boot.multicm.patch)
Following OMAP boards' methodology, I added mulitple core module configuration. For example, integratorcp_946es_config is for IntegratorCP board with ARM946ES, integratorcp_config is for IntegratorCP board with ARM926EJS by default, integratorap_946es_config is for IntegratorAP board with ARM946ES, and integratorap_config is for IntegratorAP with ARM926EJS. Now only two core modules are supported, ARM946ES and ARM926EJS.
ARM946ES support (u-boot.arm946es.patch)
ARM926EJS bug fix (u-boot.arm926ejs.patch)
Fixed CFG_HZ problem. Set CFG_HZ to 1000. The solution is from AT91RM9200.
- IntegratorCP bug fixes (u-boot.integratorcp.patch)
Added SDRAM size information and FLASH support.
Philippe, I don't have an ARM926EJS core module. Would you please test the CFG_HZ fix on your system? Thanks.
Dear Shawn,
in message c3d0340b0412201055558518ea@mail.gmail.com you wrote:
The patches were developed on an ARM IntegratorCP board with ARM946ES. Since the majority code of ARM946ES cpu was copied from ARM926EJS and there was bug about CFG_HZ found in ARM946ES later, I also fixed this CFG_HZ error on ARM926EJS.
The major changes are
- Mulitple core modules support (u-boot.multicm.patch)
Following OMAP boards' methodology, I added mulitple core module configuration. For example, integratorcp_946es_config is for IntegratorCP board with ARM946ES, integratorcp_config is for IntegratorCP board with ARM926EJS by default, integratorap_946es_config is for IntegratorAP board with ARM946ES, and integratorap_config is for IntegratorAP with ARM926EJS. Now only two core modules are supported, ARM946ES and ARM926EJS.
ARM946ES support (u-boot.arm946es.patch)
ARM926EJS bug fix (u-boot.arm926ejs.patch)
Fixed CFG_HZ problem. Set CFG_HZ to 1000. The solution is from AT91RM9200.
- IntegratorCP bug fixes (u-boot.integratorcp.patch)
Added SDRAM size information and FLASH support.
There is a problem with your patch. "u-boot.arm926ejs.patch" is not a patch but a tarball containing the other patches a second time. Please resubmit:
-> file u-boot.arm926ejs.patch u-boot.arm926ejs.patch: GNU tar archive -> tar tvf u-boot.arm926ejs.patch -rw-r--r-- xjin/redswitch 25849 2004-12-20 19:31:01 u-boot.arm946es.patch -rw-r--r-- xjin/redswitch 13015 2004-12-20 19:30:58 u-boot.integratorcp.patch -rw-r--r-- xjin/redswitch 8651 2004-12-20 19:31:09 u-boot.multicm.patch
Best regards,
Wolfgang Denk
Dear Wolfgang,
Sorry for the trouble.
On 5/30/05, Wolfgang Denk wd@denx.de wrote:
There is a problem with your patch. "u-boot.arm926ejs.patch" is not a patch but a tarball containing the other patches a second time. Please resubmit:
-> file u-boot.arm926ejs.patch u-boot.arm926ejs.patch: GNU tar archive -> tar tvf u-boot.arm926ejs.patch -rw-r--r-- xjin/redswitch 25849 2004-12-20 19:31:01 u-boot.arm946es.patch -rw-r--r-- xjin/redswitch 13015 2004-12-20 19:30:58 u-boot.integratorcp.patch -rw-r--r-- xjin/redswitch 8651 2004-12-20 19:31:09 u-boot.multicm.patch
Please find the correct one I submitted on Jan 20 2005 with the subject of "[PATCH] Revisted patches for ARM IntegratorCP, ARM926EJS, and ARM946ES".
You must have a long list of unmerged patches. Please be aware that Peter Pearse of ARM submitted a related patch on Feb 2 2005 with the subject of "[Patch]IntegratorAP & CP multi core module support", which is based on Jean-Paul Saman and my works. This can supercede my patch.
Regards, -Shawn.
Good morning (o;
Maybe the few OSX users can test this small patch to compile mkimage on OSX 10.4.x? There is no "fdatasync" in OSX Tiger:
Patch:
diff -purN u-boot.orig/tools/mkimage.c u-boot/tools/mkimage.c --- u-boot.orig/tools/mkimage.c 2005-05-31 11:09:02.000000000 +0300 +++ u-boot/tools/mkimage.c 2005-05-31 11:09:58.000000000 +0300 @@ -440,7 +440,7 @@ NXTARG: ; }
/* We're a bit of paranoid */ -#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__FreeBSD__) +#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__FreeBSD__) && !defined(__APPLE__) (void) fdatasync (ifd); #else (void) fsync (ifd); @@ -490,7 +490,7 @@ NXTARG: ; (void) munmap((void *)ptr, sbuf.st_size);
/* We're a bit of paranoid */ -#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__FreeBSD__) +#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__FreeBSD__) && !defined(__APPLE__) (void) fdatasync (ifd); #else (void) fsync (ifd);
Changelog:
* Patch by Richard Klingler, 31 May 2005: To compile mkimage on OSX 10.4.x.
best regards rick
In message <r02010500-1041-9E857FF0D1AC11D9ABAD00039387ACB6@[10.0.1.1]> you wrote:
Maybe the few OSX users can test this small patch to compile mkimage on OSX 10.4.x? There is no "fdatasync" in OSX Tiger:
Patch:
diff -purN u-boot.orig/tools/mkimage.c u-boot/tools/mkimage.c --- u-boot.orig/tools/mkimage.c 2005-05-31 11:09:02.000000000 +0300 +++ u-boot/tools/mkimage.c 2005-05-31 11:09:58.000000000 +0300 @@ -440,7 +440,7 @@ NXTARG: ; } =20 /* We're a bit of paranoid */ -#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__Fre= eBSD__) +#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__Fre= eBSD__) && !defined(__APPLE__) (void) fdatasync (ifd); #else (void) fsync (ifd); @@ -490,7 +490,7 @@ NXTARG: ; (void) munmap((void *)ptr, sbuf.st_size); =20 /* We're a bit of paranoid */ -#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__Fre= eBSD__) +#if defined(_POSIX_SYNCHRONIZED_IO) && !defined(__sun__) && !defined(__Fre= eBSD__) && !defined(__APPLE__) (void) fdatasync (ifd); #else (void) fsync (ifd);
Changelog:
- Patch by Richard Klingler, 31 May 2005: To compile mkimage on OSX 10.4.x.
I never received any test reports about this one.
Any comments?
Best regards,
Wolfgang Denk
participants (3)
-
richard＠uclinux.net -
Shawn Jin -
Wolfgang Denk