[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[lokeshjanghel91 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Lokesh Janghel  changed:

   What|Removed |Added

 CC||lokeshjanghel91 at gmail dot 
com

--- Comment #15 from Lokesh Janghel  ---
Created attachment 45639
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=45639=edit
Patch with new option for LSB & MSB

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #14 from Segher Boessenkool  ---
-msvr4-struct-return=standard and -msvr4-struct-return=compat, perhaps?

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[sirl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #13 from Franz Sirl  ---
Yes, I can do a patch for GCC-9. Any idea for the option naming? Like
-msvr4-struct-return-msb? Or should I consolidate -maix-struct-return and
-msvr4-struct-return into -maggr-return-mode={aix,svr4,svr4gnu}?

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #12 from Segher Boessenkool  ---
Yeah, but it is too late to make GCC 8.  Do you want to do the patch for GCC 9?
Ideally it goes into binutils, first.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[sirl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Franz Sirl  changed:

   What|Removed |Added

  Known to fail||3.1.1

--- Comment #11 from Franz Sirl  ---
It is wrong since the introduction of -msvr4-struct-return during the gcc-3.1
development. Before that GCC followed a draft ABI spec and returned all
aggregates in memory for ABI_V4.
A quick check reveals that a fix is probably easy to implement with a new
option (non-default for backwards compatibility), a small change in
rs6000_return_in_msb() and an additional value in rs6000_elf_file_end() for
".gnu_attribute 12".

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #10 from Segher Boessenkool  ---
I reopened PR85123 for the regression.  Let's keep this bug for purely
the SYSV ABI non-compliance.  This is not a regression (it is ancient).

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #9 from rguenther at suse dot de  ---
On Tue, 3 Apr 2018, segher at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762
> 
> --- Comment #8 from Segher Boessenkool  ---
> It is.  This PR is about us not following the ABI.  On trunk this code
> is miscompiled; there is at least one other PR for that.

So this is a [8 Regression] then?  Please fill out known-to-work/fail
for me to more easily classify PRs.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #8 from Segher Boessenkool  ---
It is.  This PR is about us not following the ABI.  On trunk this code
is miscompiled; there is at least one other PR for that.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Richard Biener  changed:

   What|Removed |Added

 CC||rguenth at gcc dot gnu.org

--- Comment #7 from Richard Biener  ---
Isn't this two different bugs?  From the description it looks like all of 4.6+
behave like this?  But the bisect points to sth that would be a 8+ regression?
Also I fail to see how this expand patch can cause this ABI issue.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Peter Bergner  changed:

   What|Removed |Added

 CC||bergner at gcc dot gnu.org

--- Comment #6 from Peter Bergner  ---
Adding a link to Alan's comment about why that patch is wrong:

  https://gcc.gnu.org/ml/gcc-patches/2018-03/msg01547.html

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Peter Bergner  changed:

   What|Removed |Added

 CC||seurer at gcc dot gnu.org

--- Comment #5 from Peter Bergner  ---
*** Bug 85123 has been marked as a duplicate of this bug. ***

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[amodra at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Alan Modra  changed:

   What|Removed |Added

 CC||tamar.christina at arm dot com

--- Comment #4 from Alan Modra  ---
Mainline regressed with svn rev 254862 (git commit e0da09896).

2017-11-17  Tamar Christina  

* expr.c (copy_blkmode_to_reg): Fix bitsize for targets
with fast unaligned access.
* doc/sourcebuild.texi (word_mode_no_slow_unalign): New.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #3 from Segher Boessenkool  ---
The problem in trunk happens at expand time already: it loads the 32-bit word
it created from stack, and then shifts it left by -8 bits.

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

--- Comment #2 from Segher Boessenkool  ---
4.6 .. 7.x behave like you say (I don't have older compilers handy).

[Bug target/84762] GCC for PowerPC32 violates the SysV ABI spec for small struct returns

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84762

Segher Boessenkool  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2018-03-22
 Ever confirmed|0   |1

--- Comment #1 from Segher Boessenkool  ---
On Linux, with trunk, with -msvr4-struct-return I get

f:
stwu 1,-32(1)
li 10,4386
li 9,51
sth 10,8(1)
stb 9,10(1)
ori 2,2,0
lwz 3,8(1)
addi 1,1,32
slwi 3,3,24
blr

which is simply incorrect (it returns only the undefined byte!) and without
the flag it generates

f:
li 8,4386
li 10,51
sth 8,0(3)
stb 10,2(3)
blr

(which is correct behaviour for this ABI, not sure if it is documented that
way though).

powerpc-eabi does the same thing, but with the other default.