Attention is currently required from: Eric Lai, Kane Chen, Paul Menzel, Pratikkumar V Prajapati, Tarun Tuli.
Subrata Banik has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/74769?usp=email )
Change subject: soc/intel/meteorlake: Add support for crashlog ......................................................................
Patch Set 24:
(1 comment)
Patchset:
PS23:
- I don’t see any redundant entries.
can you please share your AP log (similar one that i have pasted here)
``` [DEBUG] Region[0x0].Tag=0x7 offset=0x9b, size=0x400 [DEBUG] Region[0x1].Tag=0x0 offset=0xa00, size=0x280 [DEBUG] Region[0x2].Tag=0x0 offset=0x2bb0, size=0xa [DEBUG] Region[0x3].Tag=0x0 offset=0x3a00, size=0x80 [DEBUG] Region[0x4].Tag=0x1 offset=0x780, size=0x280 [DEBUG] Region[0x5].Tag=0x1 offset=0x12d8, size=0xa [DEBUG] Region[0x6].Tag=0x1 offset=0x1600, size=0x6e [DEBUG] Region[0x7].Tag=0x0 offset=0x0, size=0x0
[DEBUG] Region[0x0].Tag=0x7 offset=0x9b, size=0x400 [DEBUG] Region[0x1].Tag=0x0 offset=0xa00, size=0x280 [DEBUG] Region[0x2].Tag=0x0 offset=0x2bb0, size=0xa [DEBUG] Region[0x3].Tag=0x0 offset=0x3a00, size=0x80 [DEBUG] Region[0x4].Tag=0x1 offset=0x780, size=0x280 [DEBUG] Region[0x5].Tag=0x1 offset=0x12d8, size=0xa [DEBUG] Region[0x6].Tag=0x1 offset=0x1600, size=0x6e
```
my question is how to understand from the above entry which one is for SOC.PMC and IOE.PMC?
in crashlog.h #define CRASHLOG_DESCRIPTOR_TABLE_TAG_SOC 0x0 #define CRASHLOG_DESCRIPTOR_TABLE_TAG_IOE 0x1
We parse descriptor table to learn about crashlog records in SOC PMC, and OIE PMC SRAM. Both data has different tags. Tag=7 contain BDF info for IOE PMC SRAM device, but since we already define it in header (#define PCI_DEV_IOE_SRAM) i don’t parse the tag=7. We parse descriptor table entries with tag=0 and tag=1 only. In first pass we collect SOC PMC crashlog records and also count IOE PMC records. This is needed to allocate right size of cbmem for IOE crashlog data. After this parsing i allocate CBMEM for IOE crashlog and parse descriptor table again to collect IOE crashlog data.
- while copying data from SRAM, we ignore invalid crashlog entries. So the final size of crashlog is adjusted.
thanks
- that is coming from common code, pushed another patch.
+2'ed