-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 07/07/2011 01:16 AM, Kenneth Graunke wrote:
> On 07/05/2011 03:07 PM, Paul Berry wrote:
>> Previously, do_lower_jumps.cpp determined whether to lower return
>> statements in ir_lower_jumps_visitor::should_lower_jumps(). Moved
>> this logic to ir_lower_jumps_visitor::visit(ir_function_signature *),
>> so that it can be used in determining whether to lower a return
>> statement at the end of a function.
>> ---
>> src/glsl/lower_jumps.cpp | 19 ++++++++++++-------
>> 1 files changed, 12 insertions(+), 7 deletions(-)
>>
>> diff --git a/src/glsl/lower_jumps.cpp b/src/glsl/lower_jumps.cpp
>> index da85c6b..fa247c6 100644
>> --- a/src/glsl/lower_jumps.cpp
>> +++ b/src/glsl/lower_jumps.cpp
>> @@ -210,16 +210,17 @@ struct function_record
>> ir_function_signature* signature;
>> ir_variable* return_flag; /* used to break out of all loops and
>> then jump to the return instruction */
>> ir_variable* return_value;
>> - bool is_main;
>> + bool lower_return;
>> unsigned nesting_depth;
>>
>> - function_record(ir_function_signature* p_signature = 0)
>> + function_record(ir_function_signature* p_signature = 0,
>> + bool lower_return = false)
>> {
>> this->signature = p_signature;
>> this->return_flag = 0;
>> this->return_value = 0;
>> this->nesting_depth = 0;
>> - this->is_main = this->signature&&
>> (strcmp(this->signature->function_name(), "main") == 0);
>> + this->lower_return = lower_return;
>> }
>>
>> ir_variable* get_return_flag()
>> @@ -398,10 +399,8 @@ struct ir_lower_jumps_visitor : public
>> ir_control_flow_visitor {
>> /* never lower return at the end of a this->function */
>> if(this->function.nesting_depth == 0&&
>> ir->get_next()->is_tail_sentinel())
>> lower = false;
>> - else if (this->function.is_main)
>> - lower = lower_main_return;
>> else
>> - lower = lower_sub_return;
>> + lower = this->function.lower_return;
>> break;
>> }
>> return lower;
>> @@ -833,9 +832,15 @@ lower_continue:
>> assert(!this->function.signature);
>> assert(!this->loop.loop);
>>
>> + bool lower_return;
>> + if (strcmp(ir->function_name(), "main") == 0)
>> + lower_return = lower_main_return;
>> + else
>> + lower_return = lower_sub_return;
>
> What exactly is the difference between lowering returns in "main" and
> lowering them in other subroutines? void vs. non-void is definitely
> different, but I don't see why main is special. Looking at the code, it
> doesn't actually seem to use lower_main_return anywhere...
The difference is that some architectures (e.g., i965!) *only* want
returns in main lowered.
> (I realize this code was already there, but as long as you're cleaning
> it up, maybe that can just get removed...)
>
>> function_record saved_function = this->function;
>> loop_record saved_loop = this->loop;
>> - this->function = function_record(ir);
>> + this->function = function_record(ir, lower_return);
>> this->loop = loop_record(ir);
>>
>> assert(!this->loop.loop);
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
iEYEARECAAYFAk4Xh4wACgkQX1gOwKyEAw8R1gCcDqpk/pjz2TDyNhx/NPiTSpTS
1G0AoJb5fvKmQyiBoQz0fwmdxEoN0sIV
=0/1Z
-----END PGP SIGNATURE-----
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
