On 24/06/13 12:33, Graeme Gregory wrote:
On 24/06/13 05:00, Hanjun Guo wrote:
On 2013-6-23 17:04, Graeme Gregory wrote:
On 23/06/13 06:52, Hanjun Guo wrote:
On 2013-6-22 8:01, Graeme Gregory wrote:
Hi,
I have managed to boot our ACPI enabled kernel on the beaglebone device which has working earlyprintk.
You will see from the log below there is a very familiar looking RCU stall, same place as I was seeing with Hanjun's patches. I did not check the size
Yes, I'm sure the RCU stall happened in the same place. Graeme, do you want me to check the size of the DSDT table?
Yes please, if its over approx 4000 bytes then I think its the same issue and not an issue with CPU topology.
I rebuilt the ACPI tables with I2C device and processor device in DSDT table which you send me before, and reboot the armv8 foundation model, it shows in the log:
CPU: AArch64 Processor [410fd000] revision 0 Machine: Foundation-v8A acpi: start info is 0x0000000088100000, 4300 bytes // so the size is 4300 bytes // and it is more than 4KB. acpi: sig is "ACPI" acpi: info is 00 00 10 c4 acpi: first table is "RSD " (I) entering acpi_tb_parse_root_table ACPI: RSDP 0000000088100008 00024 (v02 LINARO) ACPI: XSDT 000000008810002c 000C4 (v01 LINARO FOUNDATI 00000014 INTL 20130214) ACPI: FACP 00000000881000f0 0010C (v05 LINARO FOUNDATI 00000000 INTL 20130214) ACPI: DSDT 00000000881001fc 0024A (v01 REDHAT ARNDALE 00000002 INTL 20130214) ACPI: MSCT 0000000088100486 00090 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: MCHI 0000000088100516 00045 (v01 LINARO FOUNDATI 02000715 INTL 20130214) ACPI: FPDT 000000008810055b 00064 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: ERST 00000000881005bf 00230 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: EINJ 00000000881007ef 00130 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: CPEP 000000008810091f 00034 (v01 LINARO FOUNDATI 00000000 INTL 20130214) ACPI: UEFI 0000000088100953 00036 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: SRAT 0000000088100989 00080 (v03 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: SPMI 0000000088100a09 00041 (v04 LINARO FOUNDATI 00000000 INTL 20130214) ACPI: SLIT 0000000088100a4a 001BC (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: OEM0 0000000088100c06 00024 (v01 LINARO FOUNDATI 0000000A INTL 20130214) ACPI: MPST 0000000088100c2a 000B6 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: MCFG 0000000088100ce0 0003C (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: APIC 0000000088100d1c 000F6 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: HEST 0000000088100e12 001D4 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: GTDT 0000000088100fe6 00050 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: ECDT 0000000088101036 00042 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: BERT 0000000088101078 00030 (v01 LINARO FOUNDATI 00000001 INTL 20130214) ACPI: SSDT 00000000881010a8 00024 (v02 LINARO FOUNDATI 00000001 INTL 20130214) (I) acpi_table_init call completed (I) exit acpi_boot_table_init enter early_acpi_boot_init [...]
But it boots ok on armv8 foundation model, so does this happened only on armv7?
Thanks Hanjun
Hi Hanjun,
Thanks for the info, I'm pretty sure you are seeing the same issue as me, when table is greater than approx 4k ACPI does not function.
This does not happen on 64bit, only on 32bit.
I have narrowed down the fault, it occurs somewhere in
acpi_bus_scan() from scan.c
This issue is now fixed and I have pushed the fix to the acpi-combined branch.
If others can confirm successful boot of this branch then I can look at moving this to our main development branch progressing us to 3.10 from current 3.9
Thanks
Graeme