llvm dev - Dec 2018 - Where's the optimiser gone? (part 5.b): missed tail calls, and more...

Compile the following functions with "-O3 -target i386"
(see <https://godbolt.org/z/VmKlXL>):

long long div(long long foo, long long bar)
{
    return foo / bar;
}

On the left the generated code; on the right the expected,
properly optimised code:

div: # @div
    push  ebp                     |
    mov   ebp, esp                |
    push  dword ptr [ebp + 20]    |
    push  dword ptr [ebp + 16]    |
    push  dword ptr [ebp + 12]    |
    push  dword ptr [ebp + 8]     |
    call  __divdi3                |    jmp   __divdi3
    add   esp, 16                 |
    pop   ebp                     |
    ret                           |


long long mod(long long foo, long long bar)
{
    return foo % bar;
}

mod: # @mod
    push  ebp                     |
    mov   ebp, esp                |
    push  dword ptr [ebp + 20]    |
    push  dword ptr [ebp + 16]    |
    push  dword ptr [ebp + 12]    |
    push  dword ptr [ebp + 8]     |
    call  __moddi3                |    jmp   __moddi3
    add   esp, 16                 |
    pop   ebp                     |
    ret                           |


long long mul(long long foo, long long bar)
{
    return foo * bar;
}

mul: # @mul
    push  ebp
    mov   ebp, esp
    push  esi
    mov   ecx, dword ptr [ebp + 16]
    mov   esi, dword ptr [ebp + 8]
    mov   eax, ecx
    imul  ecx, dword ptr [ebp + 12]
    mul   esi
    imul  esi, dword ptr [ebp + 20]
    add   edx, ecx
    add   edx, esi
    pop   esi
    pop   ebp
    ret

Clang's -target option is supposed to take a cpu type and an operating
system. So "-target i386" is giving it no operatiing system. This is
preventing frame pointer elimination which is why ebp is being updated. If
you pass "-target i386-linux" you get sightly better code.

The division/remainder operations are turned into library calls as part of
instruction selection. This code is somewhat independent of how other calls
are handled. We probably don't support tail calls in it. Is it really
realistic that a user would have a non-inlined function that contains just
a division? Why should we optimize for that case?

~Craig


On Sat, Dec 1, 2018 at 9:37 AM Stefan Kanthak via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> Compile the following functions with "-O3 -target i386"
> (see <https://godbolt.org/z/VmKlXL>):
>
> long long div(long long foo, long long bar)
> {
>     return foo / bar;
> }
>
> On the left the generated code; on the right the expected,
> properly optimised code:
>
> div: # @div
>     push  ebp                     |
>     mov   ebp, esp                |
>     push  dword ptr [ebp + 20]    |
>     push  dword ptr [ebp + 16]    |
>     push  dword ptr [ebp + 12]    |
>     push  dword ptr [ebp + 8]     |
>     call  __divdi3                |    jmp   __divdi3
>     add   esp, 16                 |
>     pop   ebp                     |
>     ret                           |
>
>
> long long mod(long long foo, long long bar)
> {
>     return foo % bar;
> }
>
> mod: # @mod
>     push  ebp                     |
>     mov   ebp, esp                |
>     push  dword ptr [ebp + 20]    |
>     push  dword ptr [ebp + 16]    |
>     push  dword ptr [ebp + 12]    |
>     push  dword ptr [ebp + 8]     |
>     call  __moddi3                |    jmp   __moddi3
>     add   esp, 16                 |
>     pop   ebp                     |
>     ret                           |
>
>
> long long mul(long long foo, long long bar)
> {
>     return foo * bar;
> }
>
> mul: # @mul
>     push  ebp
>     mov   ebp, esp
>     push  esi
>     mov   ecx, dword ptr [ebp + 16]
>     mov   esi, dword ptr [ebp + 8]
>     mov   eax, ecx
>     imul  ecx, dword ptr [ebp + 12]
>     mul   esi
>     imul  esi, dword ptr [ebp + 20]
>     add   edx, ecx
>     add   edx, esi
>     pop   esi
>     pop   ebp
>     ret
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20181201/08a35b72/attachment-0001.html>

"Craig Topper" <craig.topper at gmail.com> wrote:

> Clang's -target option is supposed to take a cpu type and an operating
> system. So "-target i386" is giving it no operatiing system. This
is
> preventing frame pointer elimination which is why ebp is being updated. If
> you pass "-target i386-linux" you get sightly better code.
The frame pointer is but not the point here.
> The division/remainder operations are turned into library calls as part of
> instruction selection. This code is somewhat independent of how other calls
> are handled. We probably don't support tail calls in it. Is it really
> realistic that a user would have a non-inlined function that contains just
> a division? Why should we optimize for that case?
I've seen quite some libraries which implement such functions, calling
just another function having the same prototype, as target-independent
wrappers.
So the question is not whether it's just a division, but in general the
call of a function having the same prototype.

regards
Stefan
> On Sat, Dec 1, 2018 at 9:37 AM Stefan Kanthak via llvm-dev <
> llvm-dev at lists.llvm.org> wrote:
> 
>> Compile the following functions with "-O3 -target i386"
>> (see <https://godbolt.org/z/VmKlXL>):
>>
>> long long div(long long foo, long long bar)
>> {
>>     return foo / bar;
>> }
>>
>> On the left the generated code; on the right the expected,
>> properly optimised code:
>>
>> div: # @div
>>     push  ebp                     |
>>     mov   ebp, esp                |
>>     push  dword ptr [ebp + 20]    |
>>     push  dword ptr [ebp + 16]    |
>>     push  dword ptr [ebp + 12]    |
>>     push  dword ptr [ebp + 8]     |
>>     call  __divdi3                |    jmp   __divdi3
>>     add   esp, 16                 |
>>     pop   ebp                     |
>>     ret                           |
>>
>>
>> long long mod(long long foo, long long bar)
>> {
>>     return foo % bar;
>> }
>>
>> mod: # @mod
>>     push  ebp                     |
>>     mov   ebp, esp                |
>>     push  dword ptr [ebp + 20]    |
>>     push  dword ptr [ebp + 16]    |
>>     push  dword ptr [ebp + 12]    |
>>     push  dword ptr [ebp + 8]     |
>>     call  __moddi3                |    jmp   __moddi3
>>     add   esp, 16                 |
>>     pop   ebp                     |
>>     ret                           |
>>
>>
>> long long mul(long long foo, long long bar)
>> {
>>     return foo * bar;
>> }
>>
>> mul: # @mul
>>     push  ebp
>>     mov   ebp, esp
>>     push  esi
>>     mov   ecx, dword ptr [ebp + 16]
>>     mov   esi, dword ptr [ebp + 8]
>>     mov   eax, ecx
>>     imul  ecx, dword ptr [ebp + 12]
>>     mul   esi
>>     imul  esi, dword ptr [ebp + 20]
>>     add   edx, ecx
>>     add   edx, esi
>>     pop   esi
>>     pop   ebp
>>     ret
>> _______________________________________________
>> LLVM Developers mailing list
>> llvm-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>
>