Re: [PATCH RFC 4/5] acpi/generic_event_device: add logic to detect if HEST addr is available

From: Mauro Carvalho Chehab
Date: Tue Nov 12 2024 - 13:04:47 EST


Em Tue, 12 Nov 2024 15:55:57 +0100
Mauro Carvalho Chehab <mchehab+huawei@xxxxxxxxxx> escreveu:

> Em Thu, 3 Oct 2024 16:27:28 +0200
> Igor Mammedov <imammedo@xxxxxxxxxx> escreveu:
>
> > > +++ b/hw/acpi/ghes.c
> > > @@ -513,7 +513,7 @@ void ghes_record_cper_errors(const void *cper, size_t len,
> > > }
> > > ags = &acpi_ged_state->ghes_state;
> > >
> > > - if (!ags->hest_addr_le) {
> > > + if (!ags->hest_lookup) {
> > > get_ghes_offsets(le64_to_cpu(ags->hw_error_le),
> > > &cper_addr, &read_ack_register_addr);
> >
> > just fencing off lookup is not enough,
> > to be compatible with qemu-9.1 (virt-9.1) we also should not publish hest_addr fwcfg.
>
> I tried this:
>
> diff --git a/hw/acpi/ghes.c b/hw/acpi/ghes.c
> index 201e72516608..6bb962d3c449 100644
> --- a/hw/acpi/ghes.c
> +++ b/hw/acpi/ghes.c
> @@ -402,8 +402,10 @@ void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s,
> fw_cfg_add_file_callback(s, ACPI_HW_ERROR_ADDR_FW_CFG_FILE, NULL, NULL,
> NULL, &(ags->hw_error_le), sizeof(ags->hw_error_le), false);
>
> - fw_cfg_add_file_callback(s, ACPI_HEST_ADDR_FW_CFG_FILE, NULL, NULL,
> - NULL, &(ags->hest_addr_le), sizeof(ags->hest_addr_le), false);
> + if (ags->hest_lookup) {
> + fw_cfg_add_file_callback(s, ACPI_HEST_ADDR_FW_CFG_FILE, NULL, NULL,
> + NULL, &(ags->hest_addr_le), sizeof(ags->hest_addr_le), false);
> + }
>
> ags->present = true;
> }
>
> But with such change, boot fails:
>
> EFI stub: Booting Linux Kernel...
> UpdateRegionMappingRecursive(0): DF100000 - E1B90000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(1): DF100000 - E1B90000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(2): C0000000 - 100000000 set 6000000000070C clr 0
> UpdateRegionMappingRecursive(2): DF100000 - E1B90000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(3): DF000000 - DF200000 set 6000000000070C clr 0
> UpdateRegionMappingRecursive(3): DF100000 - DF200000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(3): E1A00000 - E1C00000 set 6000000000070C clr 0
> UpdateRegionMappingRecursive(3): E1A00000 - E1B90000 set 400 clr FF9F000000000B3F
> EFI stub: Generating empty DTB
> EFI stub: Exiting boot services...
> UpdateRegionMappingRecursive(0): 139AC1000 - 139CD0000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(1): 139AC1000 - 139CD0000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(2): 139AC1000 - 139CD0000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(3): 139A00000 - 139C00000 set 6000000000070C clr 0
> UpdateRegionMappingRecursive(3): 139AC1000 - 139C00000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(3): 139C00000 - 139CD0000 set 400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(0): 139AC1000 - 139AD0000 set 60000000000400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(1): 139AC1000 - 139AD0000 set 60000000000400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(2): 139AC1000 - 139AD0000 set 60000000000400 clr FF9F000000000B3F
> UpdateRegionMappingRecursive(3): 139AC1000 - 139AD0000 set 60000000000400 clr FF9F000000000B3F
> SetUefiImageMemoryAttributes - 0x000000013FE60000 - 0x0000000000040000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13FE60000 - 13FEA0000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13FE60000 - 13FEA0000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13FE60000 - 13FEA0000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13FE60000 - 13FEA0000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013CAF0000 - 0x0000000000040000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13CAF0000 - 13CB30000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13CAF0000 - 13CB30000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13CAF0000 - 13CB30000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13CAF0000 - 13CB30000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013CAA0000 - 0x0000000000040000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13CAA0000 - 13CAE0000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13CAA0000 - 13CAE0000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13CAA0000 - 13CAE0000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13CAA0000 - 13CAE0000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013CA50000 - 0x0000000000040000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13CA50000 - 13CA90000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13CA50000 - 13CA90000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13CA50000 - 13CA90000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13CA50000 - 13CA90000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013C960000 - 0x0000000000040000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13C960000 - 13C9A0000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13C960000 - 13C9A0000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13C960000 - 13C9A0000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13C960000 - 13C9A0000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013FE20000 - 0x0000000000030000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13FE20000 - 13FE50000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13FE20000 - 13FE50000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13FE20000 - 13FE50000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13FE20000 - 13FE50000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013C7B0000 - 0x0000000000030000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13C7B0000 - 13C7E0000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13C7B0000 - 13C7E0000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13C7B0000 - 13C7E0000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13C7B0000 - 13C7E0000 set 70C clr 0
> SetUefiImageMemoryAttributes - 0x000000013C770000 - 0x0000000000030000 (0x0000000000000008)
> UpdateRegionMappingRecursive(0): 13C770000 - 13C7A0000 set 70C clr 0
> UpdateRegionMappingRecursive(1): 13C770000 - 13C7A0000 set 70C clr 0
> UpdateRegionMappingRecursive(2): 13C770000 - 13C7A0000 set 70C clr 0
> UpdateRegionMappingRecursive(3): 13C770000 - 13C7A0000 set 70C clr 0
>
> At this point, nothing else appears, and bios doesn't boot OSPM.
>
> (I'm using an arm64 BIOS with debug enabled)
>
> Thanks,
> Mauro

Got it. In order to be able to remove a call to
fw_cfg_add_file_callback(), no calls to bios_linker_loader_write_pointer()
can happen.

That basically explains why we can't do:

if (!ags->hest_lookup) {
fw_cfg_add_file_callback(s, ACPI_HW_ERROR_ADDR_FW_CFG_FILE, NULL, NULL,
NULL, &(ags->hw_error_le), sizeof(ags->hw_error_le), false);
}

We need the BIOS file callback to solve all the pointers that
were created between HEST table and the hardware error table.

This hunk worked:


diff --git a/hw/acpi/ghes.c b/hw/acpi/ghes.c
index 201e72516608..245efde75a8f 100644
--- a/hw/acpi/ghes.c
+++ b/hw/acpi/ghes.c
@@ -385,10 +385,12 @@ void acpi_build_hest(GArray *table_data, GArray *hardware_errors,
* tell firmware to write into GPA the address of HEST via fw_cfg,
* once initialized.
*/
- bios_linker_loader_write_pointer(linker,
- ACPI_HEST_ADDR_FW_CFG_FILE, 0,
- sizeof(uint64_t),
- ACPI_BUILD_TABLE_FILE, hest_offset);
+ if (ags->hest_lookup) {
+ bios_linker_loader_write_pointer(linker,
+ ACPI_HEST_ADDR_FW_CFG_FILE, 0,
+ sizeof(uint64_t),
+ ACPI_BUILD_TABLE_FILE, hest_offset);
+ }
}

void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s,
@@ -402,8 +404,10 @@ void acpi_ghes_add_fw_cfg(AcpiGhesState *ags, FWCfgState *s,
fw_cfg_add_file_callback(s, ACPI_HW_ERROR_ADDR_FW_CFG_FILE, NULL, NULL,
NULL, &(ags->hw_error_le), sizeof(ags->hw_error_le), false);

- fw_cfg_add_file_callback(s, ACPI_HEST_ADDR_FW_CFG_FILE, NULL, NULL,
- NULL, &(ags->hest_addr_le), sizeof(ags->hest_addr_le), false);
+ if (ags->hest_lookup) {
+ fw_cfg_add_file_callback(s, ACPI_HEST_ADDR_FW_CFG_FILE, NULL, NULL,
+ NULL, &(ags->hest_addr_le), sizeof(ags->hest_addr_le), false);
+ }

ags->present = true;
}

Thanks,
Mauro