acpi/prmt: find block with specific type
PATCH 1/1 acpi/prmt: find block with specific type
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type EFI_CONVENTIONAL_MEMORY, which is not appropriate for runtime services as described in Section 2.2.2 (Runtime Services) of the UEFI Specification [1]. Since the PRM handler is a type of runtime service, this causes an exception when the PRM handler is called.
[Firmware Bug]: Unable to handle paging request in EFI runtime service
WARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime-wrappers.c:341 __efi_queue_work+0x11c/0x170
Call trace:
__efi_queue_work+0x11c/0x170
efi_call_acpi_prm_handler+0x68/0xd0
acpi_platformrt_space_handler+0x198/0x258
acpi_ev_address_space_dispatch+0x144/0x388
acpi_ex_access_region+0x9c/0x118
acpi_ex_write_serial_bus+0xc4/0x218
acpi_ex_write_data_to_field+0x168/0x218
acpi_ex_store_object_to_node+0x1a8/0x258
acpi_ex_store+0xec/0x330
acpi_ex_opcode_1A_1T_1R+0x15c/0x618
acpi_ds_exec_end_op+0x274/0x548
acpi_ps_parse_loop+0x10c/0x6b8
acpi_ps_parse_aml+0x140/0x3b0
acpi_ps_execute_method+0x12c/0x2a0
acpi_ns_evaluate+0x210/0x310
acpi_evaluate_object+0x178/0x358
acpi_proc_write+0x1a8/0x8a0 [acpi_call]
proc_reg_write+0xcc/0x150
vfs_write+0xd8/0x380
ksys_write+0x70/0x120
__arm64_sys_write+0x24/0x48
invoke_syscall.constprop.0+0x80/0xf8
do_el0_svc+0x50/0x110
el0_svc+0x48/0x1d0
el0t_64_sync_handler+0x15c/0x178
el0t_64_sync+0x1a8/0x1b0Find a block with specific type to fix this.
prmt find a block with EFI_RUNTIME_SERVICES_DATA for prm handler. prmt find a block with EFI_RUNTIME_SERVICES_CODE for prm context. By using the correct memory types for runtime services,
we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during runtime, preventing the paging request error.
[1] https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdf
--- drivers/acpi/prmt.c | 46 ++++++++++++++++++++++++++++++--------------- 1 file changed, 31 insertions(+), 15 deletions(-)
drivers/acpi/prmt.c +31 -15
@@ -72,17 +72,21 @@ struct prm_module_info { struct prm_handler_info handlers[] __counted_by(handler_count); }; -static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type) {efi_memory_desc_t *md;u64 pa_offset = pa & ~PAGE_MASK ;u64 page = pa &PAGE_MASK ;for_each_efi_memory_desc (md) { -if (md->phys_addr < pa && pa < md->phys_addr +PAGE_SIZE * md->num_pages) +if ((md->type == type) && + (md->phys_addr < pa && pa < md->phys_addr +PAGE_SIZE * md->num_pages)) {return pa_offset + md->virt_addr + page - md->phys_addr; + } } +pr_err ("PRM: Failed to find a block for pa: %lx type %u\n" , pa, type); +return 0 ; }
@@ -148,9 +152,12 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end) th = &tm->handlers[cur_handler];guid_copy (&th->guid, (guid_t *)handler_info->handler_guid); - th->handler_addr = (void *)efi_pa_va_lookup (handler_info->handler_address); - th->static_data_buffer_addr =efi_pa_va_lookup (handler_info->static_data_buffer_address); - th->acpi_param_buffer_addr =efi_pa_va_lookup (handler_info->acpi_param_buffer_address); + th->handler_addr = + (void *)efi_pa_va_lookup (handler_info->handler_address,EFI_RUNTIME_SERVICES_CODE ); + th->static_data_buffer_addr = +efi_pa_va_lookup (handler_info->static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA ); + th->acpi_param_buffer_addr = +efi_pa_va_lookup (handler_info->acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA ); }while (++cur_handler < tm->handler_count && (handler_info =get_next_handler (handler_info)));return 0 ;
@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function, handler =find_prm_handler (&buffer->handler_guid); module =find_prm_module (&buffer->handler_guid); -if (!handler || !module) - goto invalid_guid; +if (!handler || !module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + } + +if (!handler->handler_addr || !handler->static_data_buffer_addr || + !handler->acpi_param_buffer_addr) { + buffer->prm_status =PRM_HANDLER_ERROR ; + goto error; + }ACPI_COPY_NAMESEG (context.signature,"PRMC" ); context.revision =0x0 ;
@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case PRM_CMD_START_TRANSACTION : module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) module->updatable = false;
@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case PRM_CMD_END_TRANSACTION : module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) buffer->prm_status =UPDATE_UNLOCK_WITHOUT_LOCK ;
@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function,break ; } -return AE_OK ; - -invalid_guid: - buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; +error:return AE_OK ; }
Signature · 1 line
2.43.0
Quoted 11 lines · KobaK
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type
EFI_CONVENTIONAL_MEMORY,
which is not appropriate for runtime services as described in Section 2.2.2 (Runtime
Services) of the UEFI Specification [1]. Since the PRM handler is a type of runtime
service, this causes an exception when the PRM handler is called.
Too many characters in one line.
https://docs.kernel.org/process/submitting-patches.html#the-canonical-patch-format
Quoted 32 lines
[Firmware Bug]: Unable to handle paging request in EFI runtime service
WARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime- wrappers.c:341 __efi_queue_work+0x11c/0x170 Call trace: __efi_queue_work+0x11c/0x170 efi_call_acpi_prm_handler+0x68/0xd0 acpi_platformrt_space_handler+0x198/0x258 acpi_ev_address_space_dispatch+0x144/0x388 acpi_ex_access_region+0x9c/0x118 acpi_ex_write_serial_bus+0xc4/0x218 acpi_ex_write_data_to_field+0x168/0x218 acpi_ex_store_object_to_node+0x1a8/0x258 acpi_ex_store+0xec/0x330 acpi_ex_opcode_1A_1T_1R+0x15c/0x618 acpi_ds_exec_end_op+0x274/0x548 acpi_ps_parse_loop+0x10c/0x6b8 acpi_ps_parse_aml+0x140/0x3b0 acpi_ps_execute_method+0x12c/0x2a0 acpi_ns_evaluate+0x210/0x310 acpi_evaluate_object+0x178/0x358 acpi_proc_write+0x1a8/0x8a0 [acpi_call] proc_reg_write+0xcc/0x150 vfs_write+0xd8/0x380 ksys_write+0x70/0x120 __arm64_sys_write+0x24/0x48 invoke_syscall.constprop.0+0x80/0xf8 do_el0_svc+0x50/0x110 el0_svc+0x48/0x1d0 el0t_64_sync_handler+0x15c/0x178 el0t_64_sync+0x1a8/0x1b0
Quoted 8 lines
Find a block with specific type to fix this.
prmt find a block withEFI_RUNTIME_SERVICES_DATAfor prm handler. prmt find a block withEFI_RUNTIME_SERVICES_CODEfor prm context. By using the correct memory types for runtime services, we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during runtime,
preventing the paging request error.
some general rules to follow when writing a changelog
https://docs.kernel.org/process/maintainer-tip.html 4.2.3. Changelog
Quoted 36 lines
[1]
https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdfFixes cefc7ca46235 ("ACPI: PRM: implement OperationRegion handlerfor the PlatformRtMechanism subtype")
Signed-off-by KobaK <[email protected]>Reviewed-by Matthew R. Ochs <[email protected]>---
drivers/acpi/prmt.c | 46 ++++++++++++++++++++++++++++++------------- --1 file changed, 31 insertions(+), 15 deletions(-)drivers/acpi/prmt.c +0
-0@@ -72,17 +72,21 @@ struct prm_module_info {struct
prm_handler_infohandlers[]__counted_by(handler_count);
};drivers/acpi/prmt.c +1
-1-static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type){
efi_memory_desc_t *md;
u64pa_offset= pa & ~PAGE_MASK;
u64 page = pa &PAGE_MASK;
for_each_efi_memory_desc(md) {
- if (md->phys_addr< pa && pa <md->phys_addr+PAGE_SIZE*md->num_pages)
+ if ((md->type== type) &&
+ (md->phys_addr < pa && pa <md->phys_addr+PAGE_SIZE*md->num_pages)) {
returnpa_offset+md->virt_addr+ page - md-
Quoted 7 lines
Quoted 1 line
phys_addr;+ }
}
+ pr_err("PRM: Failed to find a block for pa: %lx type %u\n", pa, type);
+
If it is a pr_err, why not error out?
or what is the proper handling for such failures?
Quoted 31 lines
return 0;
}drivers/acpi/prmt.c +0
-0@@ -148,9 +152,12 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end) th = &tm->handlers[cur_handler];
guid_copy(&th->guid, (guid_t*)handler_info-Quoted 1 line
handler_guid);- th->handler_addr = (void *)efi_pa_va_lookup(handler_info->handler_address);
- th->static_data_buffer_addr =efi_pa_va_lookup(handler_info->static_data_buffer_address); - th->acpi_param_buffer_addr =efi_pa_va_lookup(handler_info->acpi_param_buffer_address); + th->handler_addr =
+ (void *)efi_pa_va_lookup(handler_info-Quoted 1 line
handler_address,EFI_RUNTIME_SERVICES_CODE);+ th->static_data_buffer_addr = + efi_pa_va_lookup(handler_info-
Quoted 1 line
static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA);+ th->acpi_param_buffer_addr = + efi_pa_va_lookup(handler_info-
Quoted 1 line
acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA);} while (++cur_handler <
tm->handler_count&& (handler_info=get_next_handler(handler_info)));
return 0;drivers/acpi/prmt.c +0
-0@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function,
Quoted 8 lines
handler =
find_prm_handler(&buffer->handler_guid); module =find_prm_module(&buffer->handler_guid); - if (!handler || !module)
- gotoinvalid_guid;
+ if (!handler || !module) {
+ buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
I think it is okay to return AE_OK directly, right?
thanks,
rui
Quoted 55 lines
+ }
+
+ if (!handler->handler_addr || !handler-Quoted 1 line
static_data_buffer_addr||+ !handler->acpi_param_buffer_addr) { + buffer->prm_status =
PRM_HANDLER_ERROR; + goto error;
+ }
ACPI_COPY_NAMESEG(context.signature, "PRMC"); context.revision =0x0;drivers/acpi/prmt.c +0
-0@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_START_TRANSACTION:
module =find_prm_module(&buffer->handler_guid); - if (!module)
- gotoinvalid_guid;
+ if (!module) {
+ buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }
if (module->updatable)module->updatable= false;drivers/acpi/prmt.c +0
-0@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_END_TRANSACTION:
module =find_prm_module(&buffer->handler_guid); - if (!module)
- gotoinvalid_guid;
+ if (!module) {
+ buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }
if (module->updatable)buffer->prm_status=UPDATE_UNLOCK_WITHOUT_LOCK;drivers/acpi/prmt.c +0
-0@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function, break; }
- returnAE_OK;drivers/acpi/prmt.c +1
-3- -invalid_guid: - buffer->prm_status = PRM_HANDLER_GUID_NOT_FOUND; +error:return
AE_OK;
}
Quoted 15 lines · Rui
External email: Use caution opening links or attachments
Quoted 11 lines · KobaK
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type
EFI_CONVENTIONAL_MEMORY,
which is not appropriate for runtime services as described in Section 2.2.2 (Runtime
Services) of the UEFI Specification [1]. Since the PRM handler is a type of runtime
service, this causes an exception when the PRM handler is called.
Quoted 2 lines
Too many characters in one line.
https://docs.kernel.org/process/submitting-patches.html#the-canonical-patch-format
will fix this in the description.
Quoted 42 lines
Quoted 34 lines
[Firmware Bug]: Unable to handle paging request in EFI runtime
serviceWARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime- wrappers.c:341 __efi_queue_work+0x11c/0x170 Call trace: __efi_queue_work+0x11c/0x170 efi_call_acpi_prm_handler+0x68/0xd0 acpi_platformrt_space_handler+0x198/0x258 acpi_ev_address_space_dispatch+0x144/0x388 acpi_ex_access_region+0x9c/0x118 acpi_ex_write_serial_bus+0xc4/0x218 acpi_ex_write_data_to_field+0x168/0x218 acpi_ex_store_object_to_node+0x1a8/0x258 acpi_ex_store+0xec/0x330 acpi_ex_opcode_1A_1T_1R+0x15c/0x618 acpi_ds_exec_end_op+0x274/0x548 acpi_ps_parse_loop+0x10c/0x6b8 acpi_ps_parse_aml+0x140/0x3b0 acpi_ps_execute_method+0x12c/0x2a0 acpi_ns_evaluate+0x210/0x310 acpi_evaluate_object+0x178/0x358 acpi_proc_write+0x1a8/0x8a0 [acpi_call] proc_reg_write+0xcc/0x150 vfs_write+0xd8/0x380 ksys_write+0x70/0x120 __arm64_sys_write+0x24/0x48 invoke_syscall.constprop.0+0x80/0xf8 do_el0_svc+0x50/0x110 el0_svc+0x48/0x1d0 el0t_64_sync_handler+0x15c/0x178 el0t_64_sync+0x1a8/0x1b0Quoted 8 lines
Find a block with specific type to fix this.
prmt find a block withEFI_RUNTIME_SERVICES_DATAfor prm handler. prmt find a block withEFI_RUNTIME_SERVICES_CODEfor prm context. By using the correct memory types for runtime services, we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during runtime,
preventing the paging request error.
Quoted 2 lines
some general rules to follow when writing a changelog
https://docs.kernel.org/process/maintainer-tip.html 4.2.3. Changelog
will decorate this.
Quoted 87 lines
Quoted 41 lines
[1]
https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdfFixes cefc7ca46235 ("ACPI: PRM: implement OperationRegion handlerfor the PlatformRtMechanism subtype")
Signed-off-by KobaK <[email protected]>Reviewed-by Matthew R. Ochs <[email protected]>--- drivers/acpi/prmt.c | 46 ++++++++++++++++++++++++++++++------------- -- 1 file changed, 31 insertions(+), 15 deletions(-)drivers/acpi/prmt.c +5
-2@@ -72,17 +72,21 @@ struct prm_module_info { struct prm_handler_info handlers[]__counted_by (handler_count ); }; -static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type) {efi_memory_desc_t *md;u64 pa_offset = pa & ~PAGE_MASK ;u64 page = pa &PAGE_MASK ;for_each_efi_memory_desc (md) { -if (md->phys_addr < pa && pa < md->phys_addr + PAGE_SIZE * md->num_pages) +if ((md->type == type) && + (md->phys_addr < pa && pa < md->phys_addr +PAGE_SIZE * md->num_pages)) {return pa_offset + md->virt_addr + page - md- > phys_addr; + } } + pr_err("PRM: Failed to find a block for pa: %lx type %u\n",Quoted 2 lines
pa, type);
+If it is a
pr_err, why not error out?
or what is the proper handling for such failures?Quoted 29 lines
return 0;}
drivers/acpi/prmt.c +0
-3@@ -148,9 +152,12 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end) th = &tm->handlers[cur_handler];guid_copy (&th->guid, (guid_t *)handler_info- > handler_guid); - th->handler_addr = (void *)efi_pa_va_lookup(handler_info->handler_address); - th->static_data_buffer_addr =efi_pa_va_lookup (handler_info->static_data_buffer_address); - th->acpi_param_buffer_addr =
efi_pa_va_lookup(handler_info->acpi_param_buffer_address);drivers/acpi/prmt.c +2
-0+ th->handler_addr = + (void *)efi_pa_va_lookup(handler_info-Quoted 1 line
handler_address,EFI_RUNTIME_SERVICES_CODE);drivers/acpi/prmt.c +2
-0+ th->static_data_buffer_addr = + efi_pa_va_lookup(handler_info-Quoted 1 line
static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA);drivers/acpi/prmt.c +2
-0+ th->acpi_param_buffer_addr = + efi_pa_va_lookup(handler_info-Quoted 1 line
acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA);} while (++cur_handler < tm->handler_count && (handler_info =
get_next_handler(handler_info)));return 0;drivers/acpi/prmt.c static acpi_status · line 264
Quoted 10 lines
drivers/acpi/prmt.c +3
-2@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function, handler =find_prm_handler (&buffer->handler_guid); module =find_prm_module (&buffer->handler_guid); -if (!handler || !module) - goto invalid_guid; +if (!handler || !module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error;I think it is okay to return
AE_OKdirectly, right?
Quoted 2 lines
thanks,
rui
I'm also good for this.
I followed the convention in this block.
If change to "return", i will change all "goto error". How do you think?
Quoted 55 lines
Quoted 55 lines
drivers/acpi/prmt.c +3
-0+ } + + if (!handler->handler_addr || !handler-Quoted 1 line
static_data_buffer_addr||drivers/acpi/prmt.c +8
-2+ !handler->acpi_param_buffer_addr) { + buffer->prm_status =PRM_HANDLER_ERROR ; + goto error; + }ACPI_COPY_NAMESEG (context.signature,"PRMC" ); context.revision =0x0 ;@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function, case PRM_CMD_START_TRANSACTION: module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) module->updatable = false;drivers/acpi/prmt.c +4
-2@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function, case PRM_CMD_END_TRANSACTION: module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) buffer->prm_status =
UPDATE_UNLOCK_WITHOUT_LOCK;drivers/acpi/prmt.c +1
-4@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function, break; } -return AE_OK ; - -invalid_guid: - buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; +error:return AE_OK ; }
Quoted 65 lines · Koba Ko
Quoted 17 lines · Rui
External email: Use caution opening links or attachments
Quoted 13 lines · KobaK
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type
EFI_CONVENTIONAL_MEMORY,
which is not appropriate for runtime services as described in Section
2.2.2 (Runtime
Services) of the UEFI Specification [1]. Since the PRM handler is a
type of runtime
service, this causes an exception when the PRM handler is called.Quoted 2 lines
Too many characters in one line.
https://docs.kernel.org/process/submitting-patches.html#the-canonical-patch-formatwill fix this in the description.
Quoted 44 lines
Quoted 44 lines
[Firmware Bug]: Unable to handle paging request in EFI runtime
serviceWARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime- wrappers.c:341 __efi_queue_work+0x11c/0x170 Call trace: __efi_queue_work+0x11c/0x170 efi_call_acpi_prm_handler+0x68/0xd0 acpi_platformrt_space_handler+0x198/0x258 acpi_ev_address_space_dispatch+0x144/0x388 acpi_ex_access_region+0x9c/0x118 acpi_ex_write_serial_bus+0xc4/0x218 acpi_ex_write_data_to_field+0x168/0x218 acpi_ex_store_object_to_node+0x1a8/0x258 acpi_ex_store+0xec/0x330 acpi_ex_opcode_1A_1T_1R+0x15c/0x618 acpi_ds_exec_end_op+0x274/0x548 acpi_ps_parse_loop+0x10c/0x6b8 acpi_ps_parse_aml+0x140/0x3b0 acpi_ps_execute_method+0x12c/0x2a0 acpi_ns_evaluate+0x210/0x310 acpi_evaluate_object+0x178/0x358 acpi_proc_write+0x1a8/0x8a0 [acpi_call] proc_reg_write+0xcc/0x150 vfs_write+0xd8/0x380 ksys_write+0x70/0x120 __arm64_sys_write+0x24/0x48 invoke_syscall.constprop.0+0x80/0xf8 do_el0_svc+0x50/0x110 el0_svc+0x48/0x1d0 el0t_64_sync_handler+0x15c/0x178 el0t_64_sync+0x1a8/0x1b0Find a block with specific type to fix this.
prmt find a block withEFI_RUNTIME_SERVICES_DATAfor prm handler. prmt find a block withEFI_RUNTIME_SERVICES_CODEfor prm context. By using the correct memory types for runtime services, we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during
runtime,
preventing the paging request error.
Quoted 55 lines
Quoted 3 lines
some general rules to follow when writing a changelog
https://docs.kernel.org/process/maintainer-tip.html 4.2.3.
Changelogwill decorate this.
Quoted 51 lines
Quoted 40 lines
[1]
https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdfFixes cefc7ca46235 ("ACPI: PRM: implement OperationRegionhandler
for the PlatformRtMechanism subtype")Signed-off-by KobaK <[email protected]>Reviewed-by Matthew R. Ochs <[email protected]>---
drivers/acpi/prmt.c| 46 ++++++++++++++++++++++++++++++--------
-----
--1 file changed, 31 insertions(+), 15 deletions(-)drivers/acpi/prmt.c +0
-0@@ -72,17 +72,21 @@ struct prm_module_info {struct
prm_handler_infohandlers[]__counted_by(handler_count);
};drivers/acpi/prmt.c +1
-1-static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type){
efi_memory_desc_t*md;
u64pa_offset= pa & ~PAGE_MASK;
u64 page = pa &PAGE_MASK;
for_each_efi_memory_desc(md){
- if (md->phys_addr< pa && pa <md->phys_addr+PAGE_SIZE*md->num_pages)
+ if ((md->type== type) &&
+ (md->phys_addr< pa && pa <md->phys_addr+PAGE_SIZE*md->num_pages)) {
returnpa_offset+md->virt_addr+ page
- md-Quoted 8 lines
Quoted 1 line
phys_addr;+ }
}+ pr_err("PRM: Failed to find a block for pa: %lx type %u\n",
pa, type);
+If it is a
pr_err, why not error out?
or what is the proper handling for such failures?
Not sure if you missed this one.
It is not clear what is the expected behavior in this case. Better to describe this in the changelog as well.
Quoted 43 lines
Quoted 43 lines
Quoted 33 lines
return 0;
}drivers/acpi/prmt.c +0
-0@@ -148,9 +152,12 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end) th = &tm->handlers[cur_handler];guid_copy(&th->guid, (
guid_t*)handler_info-Quoted 1 line
handler_guid);-
th->handler_addr= (void *)efi_pa_va_lookup(handler_info->handler_address);
-th->static_data_buffer_addr=efi_pa_va_lookup(handler_info->static_data_buffer_address); -th->acpi_param_buffer_addr=efi_pa_va_lookup(handler_info->acpi_param_buffer_address); +th->handler_addr=
+ (void *)efi_pa_va_lookup(handler_info-Quoted 1 line
handler_address,EFI_RUNTIME_SERVICES_CODE);+
th->static_data_buffer_addr= + efi_pa_va_lookup(handler_info-Quoted 1 line
static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA);+
th->acpi_param_buffer_addr= + efi_pa_va_lookup(handler_info-Quoted 1 line
acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA);} while (++cur_handler <
tm->handler_count&& (handler_info=get_next_handler(handler_info)));return 0;
drivers/acpi/prmt.c +0
-0@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function,handler = find_prm_handler(&buffer-
Quoted 8 lines
Quoted 1 line
handler_guid);module =
find_prm_module(&buffer->handler_guid); - if (!handler || !module)
- gotoinvalid_guid;
+ if (!handler || !module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;I think it is okay to return
AE_OKdirectly, right?
Quoted 7 lines
Quoted 2 lines
thanks,
ruiI'm also good for this.
I followed the convention in this block.
If change to "return", i will change all "goto error". How do you think?
sounds good to me.
-rui
Quoted 57 lines
Quoted 56 lines
Quoted 56 lines
+ }
+
+ if (!handler->handler_addr || !handler-Quoted 1 line
static_data_buffer_addr||+ !handler->acpi_param_buffer_addr) { +
buffer->prm_status=PRM_HANDLER_ERROR; + goto error;
+ }
ACPI_COPY_NAMESEG(context.signature, "PRMC"); context.revision =0x0;drivers/acpi/prmt.c +0
-0@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_START_TRANSACTION:module =
find_prm_module(&buffer->handler_guid); - if (!module)
- gotoinvalid_guid;
+ if (!module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }if (
module->updatable)module->updatable= false;drivers/acpi/prmt.c +0
-0@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_END_TRANSACTION:module =
find_prm_module(&buffer->handler_guid); - if (!module)
- gotoinvalid_guid;
+ if (!module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }if (
module->updatable)
buffer->prm_status=UPDATE_UNLOCK_WITHOUT_LOCK;drivers/acpi/prmt.c +1
-4@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function, break; } - return AE_OK; - -invalid_guid: - buffer->prm_status = PRM_HANDLER_GUID_NOT_FOUND; +error: return AE_OK; }
Quoted 119 lines · Rui
External email: Use caution opening links or attachments
Quoted 115 lines · Koba Ko
Quoted 17 lines · Rui
External email: Use caution opening links or attachments
Quoted 13 lines · KobaK
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type
EFI_CONVENTIONAL_MEMORY,
which is not appropriate for runtime services as described in Section
2.2.2 (Runtime
Services) of the UEFI Specification [1]. Since the PRM handler is a
type of runtime
service, this causes an exception when the PRM handler is called.Quoted 2 lines
Too many characters in one line.
https://docs.kernel.org/process/submitting-patches.html#the-canonical-patch-formatwill fix this in the description.
Quoted 43 lines
Quoted 34 lines
[Firmware Bug]: Unable to handle paging request in EFI
runtime
serviceWARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime- wrappers.c:341 __efi_queue_work+0x11c/0x170 Call trace: __efi_queue_work+0x11c/0x170 efi_call_acpi_prm_handler+0x68/0xd0 acpi_platformrt_space_handler+0x198/0x258 acpi_ev_address_space_dispatch+0x144/0x388 acpi_ex_access_region+0x9c/0x118 acpi_ex_write_serial_bus+0xc4/0x218 acpi_ex_write_data_to_field+0x168/0x218 acpi_ex_store_object_to_node+0x1a8/0x258 acpi_ex_store+0xec/0x330 acpi_ex_opcode_1A_1T_1R+0x15c/0x618 acpi_ds_exec_end_op+0x274/0x548 acpi_ps_parse_loop+0x10c/0x6b8 acpi_ps_parse_aml+0x140/0x3b0 acpi_ps_execute_method+0x12c/0x2a0 acpi_ns_evaluate+0x210/0x310 acpi_evaluate_object+0x178/0x358 acpi_proc_write+0x1a8/0x8a0 [acpi_call] proc_reg_write+0xcc/0x150 vfs_write+0xd8/0x380 ksys_write+0x70/0x120 __arm64_sys_write+0x24/0x48 invoke_syscall.constprop.0+0x80/0xf8 do_el0_svc+0x50/0x110 el0_svc+0x48/0x1d0 el0t_64_sync_handler+0x15c/0x178 el0t_64_sync+0x1a8/0x1b0Quoted 9 lines
Find a block with specific type to fix this.
prmt find a block withEFI_RUNTIME_SERVICES_DATAfor prm handler. prmt find a block withEFI_RUNTIME_SERVICES_CODEfor prm context. By using the correct memory types for runtime services, we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during
runtime,
preventing the paging request error.Quoted 3 lines
some general rules to follow when writing a changelog
https://docs.kernel.org/process/maintainer-tip.html 4.2.3. Changelogwill decorate this.
Quoted 47 lines
Quoted 47 lines
[1]
https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdfFixes cefc7ca46235 ("ACPI: PRM: implement OperationRegionhandler
for the PlatformRtMechanism subtype")Signed-off-by KobaK <[email protected]>Reviewed-by Matthew R. Ochs <[email protected]>--- drivers/acpi/prmt.c | 46 ++++++++++++++++++++++++++++++-------- ----- -- 1 file changed, 31 insertions(+), 15 deletions(-)drivers/acpi/prmt.c +6
-3@@ -72,17 +72,21 @@ struct prm_module_info { struct prm_handler_info handlers[]__counted_by (handler_count ); }; -static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type) {efi_memory_desc_t *md;u64 pa_offset = pa & ~PAGE_MASK ;u64 page = pa &PAGE_MASK ;for_each_efi_memory_desc (md) { -if (md->phys_addr < pa && pa < md->phys_addr + PAGE_SIZE * md->num_pages) +if ((md->type == type) && + (md->phys_addr < pa && pa < md->phys_addr + PAGE_SIZE * md->num_pages)) {return pa_offset + md->virt_addr + page - md- > phys_addr; + } } + pr_err("PRM: Failed to find a block for pa: %lx type%u\n",
pa, type);
+
Quoted 6 lines
Quoted 3 lines
Quoted 3 lines
If it is a
pr_err, why not error out?
or what is the proper handling for such failures?Not sure if you missed this one.
It is not clear what is the expected behavior in this case. Better to describe this in the changelog as well.
Sorry, missed.
if get failure and return 0.
in acpi_platformrt_space_handler, it takes care to handle the null pointers.
```
+ if (!handler->handler_addr || !handler->static_data_buffer_addr ||
+ !handler->acpi_param_buffer_addr) {
+ buffer->prm_status = PRM_HANDLER_ERROR;
+ goto error;
+ }
```will also update in the description.
Quoted 108 lines
Quoted 41 lines
Quoted 41 lines
Quoted 41 lines
return 0;}
drivers/acpi/prmt.c +0
-3@@ -148,9 +152,12 @@ acpi_parse_prmt(union acpi_subtable_headers *header, const unsigned long end) th = &tm->handlers[cur_handler];guid_copy (&th->guid, (guid_t *)handler_info- > handler_guid); - th->handler_addr = (void *)efi_pa_va_lookup(handler_info->handler_address); - th->static_data_buffer_addr =efi_pa_va_lookup (handler_info->static_data_buffer_address); - th->acpi_param_buffer_addr =
efi_pa_va_lookup(handler_info->acpi_param_buffer_address);drivers/acpi/prmt.c +2
-0+ th->handler_addr = + (void *)efi_pa_va_lookup(handler_info-Quoted 1 line
handler_address,EFI_RUNTIME_SERVICES_CODE);drivers/acpi/prmt.c +2
-0+ th->static_data_buffer_addr = + efi_pa_va_lookup(handler_info-Quoted 1 line
static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA);drivers/acpi/prmt.c +2
-0+ th->acpi_param_buffer_addr = + efi_pa_va_lookup(handler_info-Quoted 1 line
acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA);} while (++cur_handler < tm->handler_count &&(
handler_info=get_next_handler(handler_info)));return 0;drivers/acpi/prmt.c +3
-2@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function, handler =find_prm_handler (&buffer- > handler_guid); module =find_prm_module (&buffer->handler_guid); -if (!handler || !module) - goto invalid_guid; +if (!handler || !module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error;Quoted 8 lines
Quoted 4 lines
I think it is okay to return
AE_OKdirectly, right?thanks,
ruiI'm also good for this.
I followed the convention in this block.
If change to "return", i will change all "goto error". How do you think?sounds good to me.
-rui
Quoted 55 lines
Quoted 55 lines
Quoted 55 lines
drivers/acpi/prmt.c +3
-0+ } + + if (!handler->handler_addr || !handler-Quoted 1 line
static_data_buffer_addr||drivers/acpi/prmt.c +8
-2+ !handler->acpi_param_buffer_addr) { + buffer->prm_status =PRM_HANDLER_ERROR ; + goto error; + }ACPI_COPY_NAMESEG (context.signature,"PRMC" ); context.revision =0x0 ;@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function, case PRM_CMD_START_TRANSACTION: module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) module->updatable = false;drivers/acpi/prmt.c +4
-2@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function, case PRM_CMD_END_TRANSACTION: module =find_prm_module (&buffer->handler_guid); -if (!module) - goto invalid_guid; +if (!module) { + buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; + goto error; + }if (module->updatable) buffer->prm_status =
UPDATE_UNLOCK_WITHOUT_LOCK;drivers/acpi/prmt.c +1
-4@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function, break; } -return AE_OK ; - -invalid_guid: - buffer->prm_status =PRM_HANDLER_GUID_NOT_FOUND ; +error:return AE_OK ; }
Quoted 150 lines · Koba Ko
Quoted 128 lines · Rui
External email: Use caution opening links or attachments
Quoted 124 lines · Koba Ko
Quoted 19 lines · Rui
External email: Use caution opening links or attachments
Quoted 15 lines · KobaK
PRMT needs to find the correct type of block to translate the PA-VA mapping for EFI runtime services.
The issue arises because the PRMT is finding a block of type
EFI_CONVENTIONAL_MEMORY,
which is not appropriate for runtime services as described in Section
2.2.2 (Runtime
Services) of the UEFI Specification [1]. Since the PRM handler is
a
type of runtime
service, this causes an exception when the PRM handler is called.Quoted 2 lines
Too many characters in one line.
https://docs.kernel.org/process/submitting-patches.html#the-canonical-patch-formatwill fix this in the description.
Quoted 46 lines
Quoted 35 lines
[Firmware Bug]: Unable to handle paging request in EFI runtime
serviceWARNING: CPU: 22 PID: 4330 at drivers/firmware/efi/runtime- wrappers.c:341 __efi_queue_work+0x11c/0x170 Call trace: __efi_queue_work+0x11c/0x170 efi_call_acpi_prm_handler+0x68/0xd0 acpi_platformrt_space_handler+0x198/0x258 acpi_ev_address_space_dispatch+0x144/0x388 acpi_ex_access_region+0x9c/0x118 acpi_ex_write_serial_bus+0xc4/0x218 acpi_ex_write_data_to_field+0x168/0x218 acpi_ex_store_object_to_node+0x1a8/0x258 acpi_ex_store+0xec/0x330 acpi_ex_opcode_1A_1T_1R+0x15c/0x618 acpi_ds_exec_end_op+0x274/0x548 acpi_ps_parse_loop+0x10c/0x6b8 acpi_ps_parse_aml+0x140/0x3b0 acpi_ps_execute_method+0x12c/0x2a0 acpi_ns_evaluate+0x210/0x310 acpi_evaluate_object+0x178/0x358 acpi_proc_write+0x1a8/0x8a0 [acpi_call] proc_reg_write+0xcc/0x150 vfs_write+0xd8/0x380 ksys_write+0x70/0x120 __arm64_sys_write+0x24/0x48 invoke_syscall.constprop.0+0x80/0xf8 do_el0_svc+0x50/0x110 el0_svc+0x48/0x1d0 el0t_64_sync_handler+0x15c/0x178 el0t_64_sync+0x1a8/0x1b0Quoted 11 lines
Find a block with specific type to fix this.
prmt find a block withEFI_RUNTIME_SERVICES_DATAfor prm handler.
prmt find a block withEFI_RUNTIME_SERVICES_CODEfor prm context.
By using the correct memory types for runtime services, we can ensure that the PRM handler and
its context are properly mapped in the virtual address space during
runtime,
preventing the paging request error.Quoted 3 lines
some general rules to follow when writing a changelog
https://docs.kernel.org/process/maintainer-tip.html 4.2.3.
Changelogwill decorate this.
Quoted 51 lines
Quoted 51 lines
[1]
https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdfFixes cefc7ca46235 ("ACPI: PRM: implement OperationRegionhandler
for the PlatformRtMechanism subtype")Signed-off-by KobaK <[email protected]>Reviewed-by Matthew R. Ochs <[email protected]>---
drivers/acpi/prmt.c| 46 ++++++++++++++++++++++++++++++---
-----
-----
--1 file changed, 31 insertions(+), 15 deletions(-)drivers/acpi/prmt.c +0
-0@@ -72,17 +72,21 @@ struct prm_module_info {struct
prm_handler_infohandlers[]__counted_by(handler_count);
};drivers/acpi/prmt.c +1
-1-static u64 efi_pa_va_lookup (u64 pa) +static u64 efi_pa_va_lookup (u64 pa,u32 type){
efi_memory_desc_t*md;
u64pa_offset= pa & ~PAGE_MASK;
u64 page = pa &PAGE_MASK;
for_each_efi_memory_desc(md){
- if (md->phys_addr< pa && pa <md->phys_addr+PAGE_SIZE*md->num_pages)
+ if ((md->type== type) &&
+ (md->phys_addr< pa && pa < md-Quoted 1 line
phys_addr+
PAGE_SIZE*md->num_pages)) {
returnpa_offset+md->virt_addr+ page
- md-Quoted 1 line
phys_addr;+ }
}+ pr_err("PRM: Failed to find a block for pa: %lx type %u\n",
pa, type);
+Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
If it is a
pr_err, why not error out?
or what is the proper handling for such failures?Not sure if you missed this one.
It is not clear what is the expected behavior in this case. Better to
describe this in the changelog as well.Sorry, missed.
if get failure and return 0.
inacpi_platformrt_space_handler, it takes care to handle the null pointers.``` + if (!handler->handler_addr || !handler->static_data_buffer_addr || + !handler->acpi_param_buffer_addr) { + buffer->prm_status = PRM_HANDLER_ERROR; + goto error; + } ```
Quoted 1 line
will also update in the description.
Yeah, but I mean pr_err() may be overkill if the driver is still functional.
thanks,
rui
Quoted 117 lines
Quoted 117 lines
Quoted 46 lines
Quoted 46 lines
Quoted 46 lines
return 0;
}drivers/acpi/prmt.c +0
-0@@ -148,9 +152,12 @@ acpi_parse_prmt(union
acpi_subtable_headers
*header, const unsigned long end)
th = &tm->handlers[cur_handler];guid_copy(&th->guid, (
guid_t*)handler_info-Quoted 1 line
handler_guid);-
th->handler_addr= (void *)efi_pa_va_lookup(handler_info->handler_address); -th->static_data_buffer_addr=efi_pa_va_lookup(handler_info->static_data_buffer_address); -th->acpi_param_buffer_addr=efi_pa_va_lookup(handler_info->acpi_param_buffer_address); +th->handler_addr=
+ (void *)efi_pa_va_lookup(handler_info-Quoted 1 line
handler_address,EFI_RUNTIME_SERVICES_CODE);+
th->static_data_buffer_addr= + efi_pa_va_lookup(handler_info-Quoted 1 line
static_data_buffer_address,EFI_RUNTIME_SERVICES_DATA);+
th->acpi_param_buffer_addr= + efi_pa_va_lookup(handler_info-Quoted 1 line
acpi_param_buffer_address,EFI_RUNTIME_SERVICES_DATA);} while (++cur_handler <
tm->handler_count&& (handler_info=get_next_handler(handler_info)));return 0;
drivers/acpi/prmt.c +0
-0@@ -250,8 +257,16 @@ static acpi_status acpi_platformrt_space_handler(u32 function,handler = find_prm_handler(&buffer-
Quoted 1 line
handler_guid);module = find_prm_module(&buffer-
Quoted 1 line
handler_guid);- if (!handler || !module)
- gotoinvalid_guid;
+ if (!handler || !module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;Quoted 8 lines
Quoted 4 lines
I think it is okay to return
AE_OKdirectly, right?thanks,
ruiI'm also good for this.
I followed the convention in this block.
If change to "return", i will change all "goto error". How do you think?sounds good to me.
-rui
Quoted 59 lines
Quoted 59 lines
Quoted 59 lines
+ }
+
+ if (!handler->handler_addr || !handler-Quoted 1 line
static_data_buffer_addr||+ !handler->acpi_param_buffer_addr) { +
buffer->prm_status=PRM_HANDLER_ERROR;
+ goto error;
+ }
ACPI_COPY_NAMESEG(context.signature, "PRMC");
context.revision =0x0;drivers/acpi/prmt.c +0
-0@@ -274,8 +289,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_START_TRANSACTION:module = find_prm_module(&buffer-
Quoted 1 line
handler_guid);- if (!module)
- gotoinvalid_guid;
+ if (!module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }if (
module->updatable)module->updatable= false;drivers/acpi/prmt.c +0
-0@@ -286,8 +303,10 @@ static acpi_status acpi_platformrt_space_handler(u32 function,case
PRM_CMD_END_TRANSACTION:module = find_prm_module(&buffer-
Quoted 1 line
handler_guid);- if (!module)
- gotoinvalid_guid;
+ if (!module) {
+buffer->prm_status=PRM_HANDLER_GUID_NOT_FOUND;
+ goto error;
+ }if (
module->updatable)buffer->prm_status=UPDATE_UNLOCK_WITHOUT_LOCK;drivers/acpi/prmt.c +1
-4@@ -301,10 +320,7 @@ static acpi_status acpi_platformrt_space_handler(u32 function, break; } - return AE_OK; - -invalid_guid: - buffer->prm_status = PRM_HANDLER_GUID_NOT_FOUND; +error: return AE_OK; }
Quoted 2 lines · Rui
Yeah, but I mean
pr_err()may be overkill if the driver is still functional.
how about replace with pr_warn?
Quoted 6 lines · Koba Ko
Quoted 2 lines · Rui
Yeah, but I mean
pr_err()may be overkill if the driver is still functional.how about replace with
pr_warn?
when it fails,
1. the address space handler still returns AE_OK (is it right?)
2. I don't see how PRM_HANDLER_GUID_NOT_FOUND prm_status is handled
So, if it is a critical error, we should fail the prmt probe immediately.
If it is not, we can let space handler returns AE_OK like you do in this patch, and in this case, even a pr_info() is sufficient IMV.
thanks,
rui
Quoted 12 lines · Rui
External email: Use caution opening links or attachments
Quoted 4 lines · Koba Ko
Quoted 2 lines · Rui
Yeah, but I mean
pr_err()may be overkill if the driver is still functional.how about replace with
pr_warn?when it fails,
1. the address space handler still returnsAE_OK(is it right?)
2. I don't see howPRM_HANDLER_GUID_NOT_FOUNDprm_statusis handled
Quoted 7 lines
So, if it is a critical error, we should fail the prmt probe immediately.
If it is not, we can let space handler returnsAE_OKlike you do in this patch, and in this case, even apr_info()is sufficient IMV.thanks,
rui
Agree with you. it's worse to determine the failure on another place. better way like yours, when get failure,
just complain and block the procedure in the scene.
also will modify in the v2.
thanks
Quoted 14 lines · Koba Ko
Quoted 12 lines · Rui
External email: Use caution opening links or attachments
Quoted 4 lines · Koba Ko
Quoted 2 lines · Rui
Yeah, but I mean
pr_err()may be overkill if the driver is still functional.how about replace with
pr_warn?when it fails,
1. the address space handler still returnsAE_OK(is it right?)
2. I don't see howPRM_HANDLER_GUID_NOT_FOUNDprm_statusis handled
Quoted 7 lines
Quoted 7 lines
So, if it is a critical error, we should fail the prmt probe immediately.
If it is not, we can let space handler returnsAE_OKlike you do in this patch, and in this case, even apr_info()is sufficient IMV.thanks,
rui
After reviewed again, I think it's a not critical error here. Even the current handler fails to get VA from efi_pa_va_lookup, the next handler still have a chance to get VA successfully. Block the procedure is overkill.
I would like to use pr_warn because pr_info may not be enough IMV. How do you think!?
Thanks
Quoted 6 lines
Agree with you. it's worse to determine the failure on another place. better way like yours, when get failure,
just complain and block the procedure in the scene.
also will modify in the v2.
thanks