As title. Log from a locally modified abuild attached.
I think this is something we did some time ago with simplifying debugging facility - trying to use printk() everywhere.
I have seen i440bx build break similarly. You don't see it in this log because I am developing a fix for it. (Working mainly on a P2B-LS, I'm directly affected :-) ) i440lx uses a similar debugging facility and theoretically is also broken, but apparently there is no 440LX mainboards in the tree.
abuild is modified as below, just to get CONFIG_DEBUG_RAM_SETUP set when doing all-targets abuild:
--8<------ Index: util/abuild/abuild =================================================================== --- util/abuild/abuild (revision 6204) +++ util/abuild/abuild (working copy) @@ -58,6 +58,9 @@ # use ccache ccache=false
+# RAM debug messages +debugramsetup=true + # stackprotect mode enabled by -ns option. stackprotect=false
@@ -197,6 +200,11 @@ echo "CONFIG_CCACHE=y" >> ${build_dir}/config.build fi
+ if [ "$debugramsetup" = "true" ]; then + printf "(debugramsetup enabled) " + echo "CONFIG_DEBUG_RAM_SETUP=y" >> ${build_dir}/config.build + fi + if [ "$scanbuild" = "true" ]; then printf "(scan-build enabled) " echo "CONFIG_SCANBUILD_ENABLE=y" >> ${build_dir}/config.build
--8<------
Above is not meant to be committed, but just in case, it is Signed-off-by: Keith Hui buurin@gmail.com
On Sun, 19 Dec 2010 21:56:58 -0500, Keith Hui buurin@gmail.com wrote:
Well, that is not good. Not good at all...
As promised. Attached is my fix:
Fully convert 440BX RAM init debugging to use printk(). Fixes build breakage with CONFIG_DEBUG_RAM_SETUP enabled.
A number of debugging messages have been tweaked to take advantage of the "new found" ability to output decimal numbers. :-D
I also need to make a small change to the dimm_size struct to fix another breakage.
Build tested with the modified abuild I used to report this bug. Boot tested on P2B-LS.
Joseph, now go fix i810. ;-)
Happy holidays Keith
Signed-off-by: Keith Hui buurin@gmail.com
On Mon, Dec 20, 2010 at 11:12 AM, Joseph Smith joe@settoplinux.org wrote:
2010/12/21 Keith Hui buurin@gmail.com:
As promised. Attached is my fix:
Copy/paste:
Index: src/northbridge/intel/i440bx/raminit.c
dimm --> DIMM: +printk(BIOS_DEBUG, "DIMM: %d.0: %02x", i, device);
On Tue, Dec 21, 2010 at 8:53 AM, Idwer Vollering vidwer@gmail.com wrote:
*snip*
Thanks. Updated in attached. I don't know what the .0 is here for so I took it out as well. Also gone is a second version of dump_spd_registers() that is #if'd out, and not too different from what's currently used.
Signed-off-by: Keith Hui buurin@gmail.com
On Tue, Dec 21, 2010 at 11:01:23PM -0500, Keith Hui wrote:
Signed-off-by: Keith Hui buurin@gmail.com
Thanks, r6206.
We should convert all CAR-using chipsets/boards to use printk() always, indeed. I might post further patches for that at some point.
Uwe.
On Mon, 20 Dec 2010 22:53:13 -0500, Keith Hui buurin@gmail.com wrote:
Joseph, now go fix i810. ;-)
Sorry, It will not be for a while.... I am to busy working on other things at the moment. Maybe Uwe, Anders, or anyone else contributing to i810?