<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Wed, Jan 21, 2015 at 2:16 PM, Chandler Carruth <span dir="ltr"><<a href="mailto:chandlerc@gmail.com" target="_blank">chandlerc@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr">So, we've run into some test cases which are pretty alarming.<div><br></div><div>When inlining code in various different paths we can end up with this IR:</div><div><br></div><div><div><font face="monospace, monospace">define void @f(float* %value, i8* %b) {<br></font></div><div><font face="monospace, monospace">entry:</font></div><div><font face="monospace, monospace"> %0 = load float* %value, align 4</font></div><div><font face="monospace, monospace"> %1 = bitcast i8* %b to float*</font></div><div><font face="monospace, monospace"> store float %0, float* %1, align 1</font></div><div><font face="monospace, monospace"> ret void</font></div><div><font face="monospace, monospace">}</font></div><div><font face="monospace, monospace"><br></font></div><div><font face="monospace, monospace">define void @g(float* %value, i8* %b) {<br></font></div><div><font face="monospace, monospace">entry:</font></div><div><font face="monospace, monospace"> %0 = bitcast float* %value to i32*</font></div><div><font face="monospace, monospace"> %1 = load i32* %0, align 4</font></div><div><font face="monospace, monospace"> %2 = bitcast i8* %b to i32*</font></div><div><font face="monospace, monospace"> store i32 %1, i32* %2, align 1</font></div><div><font face="monospace, monospace"> ret void</font></div><div><font face="monospace, monospace">}</font></div><div><br></div></div><div>Now, I don't really care one way or the other about these two IR inputs, but it's pretty concerning that we get these two equivalent bits of code and nothing canonicalizes to one or the other.</div><div><br></div><div>So, the naive first blush approach here would be to canonicalize on the first -- it has fewer instructions after all -- but I don't think that's the right approach for two reasons:</div><div><br></div><div>1) It will be a *very* narrow canonicalization that only works with overly specific sets of casted pointers.</div><div>2) It doesn't effectively move us toward the optimizer treating IR with different pointee types for pointer types indistinguishably. Some day, I continue to think we should get rid of the pointee types entirely.</div><div><br></div><div>To see why #1 and #2 are problematic, assume another round of inlining took place and we suddenly had the following IR:</div><div><br></div></div></blockquote><div><br></div><div>And the missing IR example:</div><div><br></div><div><font face="monospace, monospace">define void @f(i8* %value, i8* %b) {<br></font></div><div><font face="monospace, monospace">entry:</font></div><div><font face="monospace, monospace"> %0 = bitcast i8* %value to float*</font></div><div><font face="monospace, monospace"> %1 = load float* %0, align 4</font></div><div><font face="monospace, monospace"> %2 = bitcast i8* %b to float*</font></div><div><font face="monospace, monospace"> store float %0, float* %1, align 1</font></div><div><font face="monospace, monospace"> ret void</font></div><div><font face="monospace, monospace">}</font></div><div><font face="monospace, monospace"><br></font></div><div><font face="monospace, monospace">define void @g(i8* %value, i8* %b) {<br></font></div><div><font face="monospace, monospace">entry:</font></div><div><font face="monospace, monospace"> %0 = bitcast i8* %value to i32*</font></div><div><font face="monospace, monospace"> %1 = load i32* %0, align 4</font></div><div><font face="monospace, monospace"> %2 = bitcast i8* %b to i32*</font></div><div><font face="monospace, monospace"> store i32 %1, i32* %2, align 1</font></div><div><font face="monospace, monospace"> ret void</font></div><div><span style="font-family:monospace,monospace">}</span></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr"><div></div><div><br></div><div>AFAICT, this is the same and we still don't have a good canonicalization story.</div><div><br></div><div>What seems like the obvious important and missing canonicalization is that when we have a loaded value that is *only* used by storing it back into memory, we don't canonicalize the type of that *value* (ignoring the pointer types) to a single value type.</div><div><br></div><div>So, the only really suitable type for this kind of stuff is 'iN' where N matches the number of bits loaded or stored.</div><div><br></div><div>I have this change implemented. It is trivial and unsurprising. However, the effects of this are impossible to predict so I wanted to make sure it made sense to others. Essentially, I expect random and hard to track down performance fluctuations across the board. Some things may get better, others may get worse, and they will probably all be bugs elsewhere in the stack.</div><div><br></div><div>So, thoughts?</div><span class=""><font color="#888888"><div>-Chandler</div></font></span></div>
</blockquote></div><br></div></div>