[PATCH] drm/xe: Generate oob before compiling anything

Lucas De Marchi lucas.demarchi at intel.com
Thu Apr 25 12:48:21 UTC 2024


On Thu, Apr 25, 2024 at 02:19:45PM GMT, Jani Nikula wrote:
>On Thu, 11 Apr 2024, Lucas De Marchi <lucas.demarchi at intel.com> wrote:
>> Instead of keep adding more dependencies as WAs are needed in different
>> places of the driver, just add a rule with all the objects so the code
>> generation happens before anything else.
>>
>> Signed-off-by: Lucas De Marchi <lucas.demarchi at intel.com>
>> ---
>>  drivers/gpu/drm/xe/Makefile | 16 ++++------------
>>  1 file changed, 4 insertions(+), 12 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/xe/Makefile b/drivers/gpu/drm/xe/Makefile
>> index 6015c9e41f24..a5809b3ec8a5 100644
>> --- a/drivers/gpu/drm/xe/Makefile
>> +++ b/drivers/gpu/drm/xe/Makefile
>> @@ -46,18 +46,6 @@ $(obj)/generated/%_wa_oob.c $(obj)/generated/%_wa_oob.h: $(obj)/xe_gen_wa_oob \
>>  		 $(srctree)/$(src)/xe_wa_oob.rules
>>  	$(call cmd,wa_oob)
>>
>> -uses_generated_oob := \
>> -	$(obj)/xe_gsc.o \
>> -	$(obj)/xe_guc.o \
>> -	$(obj)/xe_guc_ads.o \
>> -	$(obj)/xe_migrate.o \
>> -	$(obj)/xe_ring_ops.o \
>> -	$(obj)/xe_vm.o \
>> -	$(obj)/xe_wa.o \
>> -	$(obj)/xe_ttm_stolen_mgr.o
>> -
>> -$(uses_generated_oob): $(generated_oob)
>> -
>>  # Please keep these build lists sorted!
>>
>>  # core driver code
>> @@ -319,3 +307,7 @@ quiet_cmd_hdrtest = HDRTEST $(patsubst %.hdrtest,%.h,$@)
>>
>>  $(obj)/%.hdrtest: $(src)/%.h FORCE
>>  	$(call if_changed_dep,hdrtest)
>> +
>> +# make sure the generated header is done before any
>> +# compilation unit is attempted
>> +$(xe-y): $(generated_oob)
>
>I'm sorry, I don't understand why this would ensure the order.

$(xe-y) should expand to every single .o listed here. So this is adding
a dependency "every single .o depends on the generated files. Or at
least it was the intention and appeared to work fine for x86, but no UM.
Before we had:

$(uses_generated_oob): $(generated_oob)

with $(uses_generated_oob) being crafted by hand.

>
>Moreover, I think there should be a way to express the dependency
>accurately and automatically. I'm just not sure how. I guess it's the
>system include <generated/...> that prevents the usual dep handling from
>working.

why? there's nothing special with "generated" or instructing the
compiler to look at the include paths (without looking at the current
dir)

Lucas De Marchi

>
>BR,
>Jani.
>
>
>-- 
>Jani Nikula, Intel


More information about the Intel-xe mailing list