RFR: 8207345: AArch64: Trampoline generation code reads from uninitialized memory

[Aleksey Shipilev]

On 07/16/2018 06:42 PM, Aleksey Shipilev wrote:
> On 07/16/2018 06:40 PM, Andrew Haley wrote:
>> On 07/16/2018 05:38 PM, Aleksey Shipilev wrote:
>>> Your choice, really. I used to frown about conditionals like this all around Hotspot, but then I
>>> started to see the appeal: you can clearly see (A && B && C) here, even if A, B, C are not trivial.
>>> But structurally, you immediately grasp that if A fails, we don't enter the block; if B fails, we
>>> don't enter the block, if C fails, we don't enter the block. It takes some time to read through
>>> chain of local variables to reach the same verdict.
>>
>> I see.  I tried it, and I think your suggestions do help, and the
>> hybrid result is best of all:
>>
>>   // We need a trampoline if branches are far.
>>   if (far_branches()) {
>>     // We don't want to emit a trampoline if C2 is generating dummy
>>     // code during its branch shortening phase.
>>     CompileTask* task = ciEnv::current()->task();
>>     if (task != NULL && is_c2_compile(task->comp_level()) &&
>>         Compile::current->in_scratch_emit_size()) {
>>       address stub = emit_trampoline_stub(offset(), entry.target());
>>       if (stub == NULL) {
>>         return NULL; // CodeCache is full
>>       }
>>     }
> 
> Looks good.

...except that it should be !Compile::current->in_scratch_emit_size()? (I made this mistake when
rewriting, sorry).

-Aleksey

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20180716/4ff046ff/signature.asc>