Linaro GCC

ICE in reload_cse_simplify_operands in thumb-1 mode

Bug #629671 reported by Loïc Minier on 2010-09-03

This bug affects 3 people

	Status	Importance	Assigned to	Milestone
Linaro GCC	Fix Released	Low	Julian Brown	Linaro GCC 4.5-2010.12-0
4.4	Fix Released	Low	Julian Brown	Linaro GCC 4.4-2010.12-0
4.5	Fix Released	Low	Julian Brown	Linaro GCC 4.5-2010.12-0
Linaro GCC Tracking	Fix Released	Undecided	Unassigned	Linaro GCC Tracking 4.6.0
gcc-4.4 (Ubuntu)	Fix Released	Medium	Unassigned
gcc-4.5 (Ubuntu)	Fix Released	Medium	Unassigned

Bug Description

as reported by Mariano Alvira <email address hidden> on linaro-dev:

$ arm-linux-gnueabi-gcc-4.5 -Os -march=armv4t -c -o tests.o tests.c
tests.c: In function 'bad':
tests.c:17:1: error: insn does not satisfy its constraints:
(insn 22 14 16 2 tests.c:14 (set (reg/f:SI 3 r3 [144])
(reg/f:SI 13 sp)) 168 {*thumb1_movsi_insn_osize} (nil))
tests.c:17:1: internal compiler error: in reload_cse_simplify_operands, at postreload.c:396
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-4.5/README.Bugs> for instructions.

(same with gcc-4.4)

this passes with -marm or with -march=armv6t2, but not with armv4t, armv5t, or armv6

attaching test case tests.c

Bye,

Tags:

Related branches

lp://staging/~julian-codesourcery/gcc-linaro/lp-629671

Merged into lp://staging/gcc-linaro/4.5 at revision 99433

Andrew Stubbs (community): Approve on 2010-11-11

lp://staging/gcc-linaro/4.5

lp://staging/~julian-codesourcery/gcc-linaro/lp-629671-4.4

Merged into lp://staging/gcc-linaro/4.4 at revision 93568

Linaro Toolchain Developers: Pending requested 2010-12-03

Revision history for this message

Loïc Minier (lool) wrote on 2010-09-03:

testcase Edit (331 bytes, text/plain)

Revision history for this message

Loïc Minier (lool) wrote on 2010-09-03:

Reproducible with hrw's gcc-4.5-arm-linux-gnueabi 4.5.1-2ubuntu2, gcc-4.4-arm-linux-gnueabi 4.4.4-8ubuntu1hrw1, and with Ubuntu's gcc-4.5 4.5.1-2ubuntu2 (has Linaro 4.5-2010.08-1) natively.

Changed in gcc-linaro:
importance:	Undecided → Low
status:	New → Confirmed

Revision history for this message

Loïc Minier (lool) wrote on 2010-09-03:

I'm putting that as Confirmed since two people see the bug and as Low since it's apparently Thumb-1 only.

Revision history for this message

Michael Hope (michaelh1) wrote on 2010-09-21:

Problem does not exist in gcc-linaro-4.5-2010.09-1 or gcc-linaro-4.4-2010.09-1.

Tracked it down a bit further. The insn that fails to match is introduced by debian/patches/armel-hilo-union-class.diff which is a fix for gfortran.dg/vector_subscript_1.f90.

Next step is to see if vector_subscript_1.f90 fails in upstream and, if not, revert the patch.

Matthias Klose (doko) on 2010-09-25

Changed in gcc-4.4 (Ubuntu):
importance:	Undecided → High
status:	New → Confirmed
importance:	High → Medium
Changed in gcc-4.5 (Ubuntu):
importance:	Undecided → Medium
status:	New → Confirmed

Michael Hope (michaelh1) on 2010-09-28

tags:

added: ice

Revision history for this message

Julian Brown (julian-codesourcery) wrote on 2010-10-19:

Fixed upstream by:

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg01142.html

Revision history for this message

Julian Brown (julian-codesourcery) wrote on 2010-10-19:

The above-mentioned patch is not yet present on the Linaro 4.5 branch.

Revision history for this message

Michael Hope (michaelh1) wrote on 2010-10-19:

Thank you. Could you please backport it to Linaro GCC 4.5. I'll decide what to do about 4.4.

Michael Hope (michaelh1) on 2010-11-19

Changed in gcc-linaro:
milestone:	none → 4.5-2010.12-0

Revision history for this message

Michael Hope (michaelh1) wrote on 2010-11-19:

Julian, could you please backport this to Linaro GCC 4.4 as well?

Revision history for this message

Levak (levak92) wrote on 2010-11-28:

Confirmed with gcc-4.5
What I have to do to fix it ?

Revision history for this message

Michael Hope (michaelh1) wrote on 2010-11-28:

#10

Hi Levak. It depends on what you want to do. I believe that the gcc-4.5 package that comes with Maverick has this bug, but the gcc that comes with Natty does not. Send me an email at michael.hope at linaro.org and we can talk further.

Revision history for this message

Julian Brown (julian-codesourcery) wrote on 2010-12-01:

#11

I'm testing the backport to Linaro GCC 4.4.

Revision history for this message

Matthias Klose (doko) wrote on 2010-12-14:

#12

fixed in 4.5.1-12ubuntu1 (Ubuntu natty)

Changed in gcc-4.5 (Ubuntu):
status:	Confirmed → Fix Released

Revision history for this message

Matthias Klose (doko) wrote on 2010-12-14:

#13

Fixed in gcc-4.4 4.4.5-10ubuntu1 (Ubuntu natty)

Changed in gcc-4.4 (Ubuntu):
status:	Confirmed → Fix Released

Revision history for this message

Andrew Stubbs (ams-codesourcery) wrote on 2011-02-10:

#14

Adding tracking for upstream submission ...

Related: lp:gcc-linaro/4.5,revno=99433

Changed in gcc-linaro-tracking:
milestone:	none → 4.7.0

Revision history for this message

Andrew Stubbs (ams-codesourcery) wrote on 2011-02-10:

#15

Oh, apparently this is a backport that didn't get labelled in the changelog. Marking as upstream.

Changed in gcc-linaro-tracking:
milestone:	4.7.0 → 4.6.0
status:	New → Fix Committed

Andrew Stubbs (ams-codesourcery) on 2011-02-25

tags:

added: 46merge

Andrew Stubbs (ams-codesourcery) on 2011-04-06

Changed in gcc-linaro-tracking:
status:	Fix Committed → Fix Released

Report a bug

This report contains Public information

Everyone can see this information.

You are

Subscribing...

Edit bug mail

Other bug subscribers

Related blueprints

Linaro GCC 4.6

Bug attachments

testcase Edit

Add attachment

Remote bug watches

Bug watches keep track of this bug in other bug trackers.