[Mesa-dev] [PATCH 03/23] intel/eu: Use brw_set_desc() along with a helper to set common descriptor controls.

[Francisco Jerez]

Kenneth Graunke <kenneth at whitecape.org> writes:

> On Monday, June 11, 2018 7:25:55 PM PDT Francisco Jerez wrote:
>> This replaces brw_set_message_descriptor() with the composition of
>> brw_set_desc() and a new inline helper function that packs the common
>> message descriptor controls into an integer.  The goal is to represent
>> all message descriptors as a 32-bit integer which is written at once
>> into the instruction, which is more flexible (SENDS anyone?), robust
>> (see d2eecf0b0b24d203d0f171807681dffd830d54de fixing an issue
>> ultimately caused by some bits of the extended message descriptor
>> being left undefined) and future-proof than the current approach of
>> specifying the individual descriptor fields directly into the
>> instruction.
>> 
>> This approach also seems more self-documenting, since it will allow
>> removing calls to functions with way too many arguments like
>> brw_set_*_message() and brw_send_indirect_message(), and instead
>> provide a single descriptor argument constructed from an appropriate
>> combination of brw_*_desc() helpers.
>> 
>> Note that because brw_set_message_descriptor() was (conditionally?)
>> overriding fields of the instruction which strictly speaking weren't
>> part of the message descriptor, this involves calling
>> brw_inst_set_sfid() and brw_inst_set_eot() in some cases in addition
>> to brw_set_desc().
>> ---
>>  src/intel/compiler/brw_eu.h               |  29 +++++---
>>  src/intel/compiler/brw_eu_emit.c          | 108 +++++++++++-------------------
>>  src/intel/compiler/brw_vec4_generator.cpp |  17 +++--
>>  3 files changed, 68 insertions(+), 86 deletions(-)
>> 
>> diff --git a/src/intel/compiler/brw_eu.h b/src/intel/compiler/brw_eu.h
>> index 5a396339fde..b2b20713e45 100644
>> --- a/src/intel/compiler/brw_eu.h
>> +++ b/src/intel/compiler/brw_eu.h
>> @@ -256,14 +256,6 @@ void brw_set_sampler_message(struct brw_codegen *p,
>>                               unsigned simd_mode,
>>                               unsigned return_format);
>>  
>> -void brw_set_message_descriptor(struct brw_codegen *p,
>> -                                brw_inst *inst,
>> -                                enum brw_message_target sfid,
>> -                                unsigned msg_length,
>> -                                unsigned response_length,
>> -                                bool header_present,
>> -                                bool end_of_thread);
>> -
>>  void brw_set_dp_read_message(struct brw_codegen *p,
>>  			     brw_inst *insn,
>>  			     unsigned binding_table_index,
>> @@ -287,6 +279,27 @@ void brw_set_dp_write_message(struct brw_codegen *p,
>>  			      unsigned end_of_thread,
>>  			      unsigned send_commit_msg);
>>  
>> +/**
>> + * Construct a message descriptor immediate with the specified common
>> + * descriptor controls.
>> + */
>> +static inline uint32_t
>> +brw_message_desc(const struct gen_device_info *devinfo,
>> +                 unsigned msg_length,
>> +                 unsigned response_length,
>> +                 bool header_present)
>> +{
>
> Perhaps it would be good to add
>
>       assert(msg_length >= 1 && msg_length <= 15);
>
>> +   if (devinfo->gen >= 5) {
>
>       assert(response_length <= 16);
>
>
>> +      return (msg_length << 25 |
>> +              response_length << 20 |
>> +              header_present << 19);
>> +   } else {
>
>       assert(response_length <= 8);
>
> I'm not so concerned with validating the values here, just thinking it
> might make sense to verify that mlen fits in a U4, for example, so we
> don't accidentally bleed over into other fields when encoding it.
>

It's kind of a PITA to assert that each field is in range manually for
each one of these helpers (the following patches introduce a pile of
functions very much like this one), and verifying that the assertions
are complete and match the definition of the hardware fields is not
straightforward to review.  I would have used the SET_FIELD() macro if
it wasn't because it relies on macros being defined with specific
suffixes for each field.  If you believe it's going to be valuable I
think I'm going to introduce a new helper more easily reusable than
SET_FIELD() checking for overflow based on a bitfield specification, and
use it instead of the plain left-shift operators.

> I suppose the validator already catches this, though...
>

I don't think the validator will be able to catch such cases in general.

>> +      return (msg_length << 20 |
>> +              response_length << 16);
>> +   }
>> +}
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 227 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20180621/ddd5e103/attachment-0001.sig>