llvm dev - Mar 2018 - Why is there no EmitInt64 in AsmPrinter?

Hi all,

The AsmPrinter class supports functions like EmitInt8, EmitInt16, and
EmitInt32 for writing a 1/2/4 byte directive to the assembly. Each of these
calls the MCStreamer::EmitIntValue method with the corresponding size. For
some reason, there is no EmitInt64, and I was wondering if there was a
fundamental reason why? The EmitIntValue function appears to support 8-byte
inputs.

I dug into this a bit more and found a commit way back in 2010 (r100296)
that removed the "EmitInt64" equivalent, with no explanation
(presumably it
was unused).

I'd find this function quite helpful - I'm currently working on
extending
the DwarfGenerator unit-test helper class to support debug line, and I'd
like to be able to write 8 byte values (e.g. to reflect DWARF64 lengths). I
could trivially reinstate the function, but if there's a good reason not
to, it would be good to know before I go to the effort of creating a review
for it.

Thanks for any help.

James
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180309/c79b6da1/attachment.html>

DWARF already needs to emit 64-bit values, being the type-unit hashes.  How is
that being done?  (If it's going directly to the streamer, then we're
already assuming all assemblers are okay with what the streamer does, and adding
an AsmPrinter helper seems very reasonable.)
--paulr

From: llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] On Behalf Of James
Henderson via llvm-dev
Sent: Friday, March 09, 2018 7:44 AM
To: llvm-dev
Subject: [llvm-dev] Why is there no EmitInt64 in AsmPrinter?

Hi all,
The AsmPrinter class supports functions like EmitInt8, EmitInt16, and EmitInt32
for writing a 1/2/4 byte directive to the assembly. Each of these calls the
MCStreamer::EmitIntValue method with the corresponding size. For some reason,
there is no EmitInt64, and I was wondering if there was a fundamental reason
why? The EmitIntValue function appears to support 8-byte inputs.
I dug into this a bit more and found a commit way back in 2010 (r100296) that
removed the "EmitInt64" equivalent, with no explanation (presumably it
was unused).
I'd find this function quite helpful - I'm currently working on
extending the DwarfGenerator unit-test helper class to support debug line, and
I'd like to be able to write 8 byte values (e.g. to reflect DWARF64
lengths). I could trivially reinstate the function, but if there's a good
reason not to, it would be good to know before I go to the effort of creating a
review for it.
Thanks for any help.
James
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180320/70ea6587/attachment-0001.html>

I took a look at the code a bit more, and it looks like there are no users
of EmitIntValue with an explicit 8-byte size, although plenty use some
other mechanism to specify the size, which could clearly be 8 bytes (e.g.
when emitting DW_FORM_data8 data, or address sizes). As far as I can tell,
there is no DWARF64 writing support anywhere in LLVM, so the obvious
use-case for writing a literal 64-bit value is not there either. Type
signatures, along with other DWARF DIEs are written with an
indirectly-specified size. I think the fact that this can be a value of 8
probably means that this is safe, but I'm uncertain.

On 20 March 2018 at 18:49, <paul.robinson at sony.com> wrote:
> DWARF already needs to emit 64-bit values, being the type-unit hashes.
> How is that being done?  (If it's going directly to the streamer, then
> we're already assuming all assemblers are okay with what the streamer
does,
> and adding an AsmPrinter helper seems very reasonable.)
>
> --paulr
>
>
>
> *From:* llvm-dev [mailto:llvm-dev-bounces at lists.llvm.org] *On Behalf Of
*James
> Henderson via llvm-dev
> *Sent:* Friday, March 09, 2018 7:44 AM
> *To:* llvm-dev
> *Subject:* [llvm-dev] Why is there no EmitInt64 in AsmPrinter?
>
>
>
> Hi all,
>
> The AsmPrinter class supports functions like EmitInt8, EmitInt16, and
> EmitInt32 for writing a 1/2/4 byte directive to the assembly. Each of these
> calls the MCStreamer::EmitIntValue method with the corresponding size. For
> some reason, there is no EmitInt64, and I was wondering if there was a
> fundamental reason why? The EmitIntValue function appears to support 8-byte
> inputs.
>
> I dug into this a bit more and found a commit way back in 2010 (r100296)
> that removed the "EmitInt64" equivalent, with no explanation
(presumably it
> was unused).
>
> I'd find this function quite helpful - I'm currently working on
extending
> the DwarfGenerator unit-test helper class to support debug line, and
I'd
> like to be able to write 8 byte values (e.g. to reflect DWARF64 lengths). I
> could trivially reinstate the function, but if there's a good reason
not
> to, it would be good to know before I go to the effort of creating a review
> for it.
>
> Thanks for any help.
>
> James
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180321/ed36b6ec/attachment.html>