Bug 60040

Created attachment 32024 [details]
Test case.

Seems like PR58545

*** This bug has been marked as a duplicate of bug 58545 ***

With avr-rtems4.11-gcc (GCC) 4.9.0 20140218 (experimental) I still get the error:

avr-rtems4.11-gcc -fpreprocessed -w -mmcu=atmega128 -O2 -s test.i -o /dev/null
test.i: In function 'rtems_fdisk_recycle_segment':
test.i:107:1: error: unable to find a register to spill in class 'POINTER_REGS'
 }
 ^
test.i:107:1: error: this is the insn:
(insn 30 29 31 2 (set (reg:HI 26 r26)
        (reg/v/f:HI 51 [ dpd ])) test.i:95 82 {*movhi}
     (nil))
test.i:107:1: internal compiler error: in spill_failure, at reload1.c:2106
0x825135 _fatal_insn(char const*, rtx_def const*, char const*, int, char const*)
        /home/sh/archive/gcc-git/gcc/rtl-error.c:109
0x81a888 spill_failure
        /home/sh/archive/gcc-git/gcc/reload1.c:2106
0x81a888 find_reload_regs
        /home/sh/archive/gcc-git/gcc/reload1.c:2032
0x8230a7 select_reload_regs
        /home/sh/archive/gcc-git/gcc/reload1.c:2052
0x8230a7 reload(rtx_def*, int)
        /home/sh/archive/gcc-git/gcc/reload1.c:973
0x72c04d do_reload
        /home/sh/archive/gcc-git/gcc/ira.c:5469
0x72c04d rest_of_handle_reload
        /home/sh/archive/gcc-git/gcc/ira.c:5598
0x72c04d execute
        /home/sh/archive/gcc-git/gcc/ira.c:5627
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

The output is different though.

Confirmed on current 4.9 trunk from 2014-02-18 (SVN 207837).

Created attachment 32372 [details]
tentative patch for tentative reloads

In this case, reload already knows that it has to re-do the reloads, but it
goes ahead anyway and computes reloads registers for this iteration.
Unfortunately, when find_reload_regs fails, it then calls spill_failure,
giving a hard error for a reload that we don't need in the first place.

The patch in this attachment passes down something_changed from reload
as tentative to select_reload_regs and then on to find_reload_regs to
not worry about the failure.
Also, in reload, I made it not 'goto failure' in that case.

GCC 4.9.0 20140407 with the proposed patch fixes the problem for me.

Was the proposed patch ever committed so this can be closed?

Seems not - just tried with avr-gcc 5.1 and it is still broken:

$ avr-gcc -fpreprocessed -w -mmcu=atmega128 -O2 -s test.i -o /dev/null
        test.i: In function 'rtems_fdisk_recycle_segment':
        test.i:107:1: error: unable to find a register to spill in class 'POINTER_REGS'
         }
         ^
        test.i:107:1: error: this is the insn:
        (insn 30 29 31 2 (set (reg:HI 26 r26)
                (reg/v/f:HI 51 [ dpd ])) /home/matthijs/test.i:95 83 {*movhi}
             (nil))
        test.i:107: confused by earlier errors, bailing out

$ avr-gcc --version
        avr-gcc (GCC) 5.1.0
        Copyright (C) 2015 Free Software Foundation, Inc.
        This is free software; see the source for copying conditions.  There is NO
        warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

On the Arduino bugtracker [1], another testcase with the same symptoms was
reported. I'm attaching that here. This testcase works with -O2, but
breaks when -Os is used.

$ avr-gcc -c -Os  -mmcu=atmega328p  test2.c -o /dev/null

        test2.c: In function 'getSlope':
        test2.c:22:1: error: unable to find a register to spill in class 'POINTER_REGS'
         }
         ^
        test2.c:22:1: error: this is the insn:
        (insn 40 38 42 3 (set (reg:SF 63 [ D.1613 ])
                (mem:SF (post_inc:HI (reg:HI 16 r16 [orig:73 ivtmp.13 ] [73])) [1 MEM[base: _27, offset: 0B]+0 S4 A8])) /home/matthijs/test.c:15 100 {*movsf}
             (expr_list:REG_INC (reg:HI 16 r16 [orig:73 ivtmp.13 ] [73])
                (nil)))
        test2.c:22: confused by earlier errors, bailing out

[1]: https://github.com/arduino/Arduino/issues/3972

Created attachment 36499 [details]
Second testcase, needs -Os to break

I've been getting this error on some complex code of mine (for atmga2560). I tried the attached patch on gcc 5.3. While it resolves the original compile error, I then get an error from the linker, an undefined reference to the same function that the original error was related to. So I think something is wrong with the patch.

Created attachment 37320 [details]
Modified patch which clears failure

I made a wild guess that the issue is that the "failure" is not cleared, so I modified the patch (by hand), to clear "failure" after an ignored failure. With this my code compiles and links.

Author: bernds
Date: Fri Apr 29 08:59:09 2016
New Revision: 235625

URL: https://gcc.gnu.org/viewcvs?rev=235625&root=gcc&view=rev
Log:
avr-related reload fix from Senthil Kumar Selvaraj

	PR target/60040
	* reload1.c (reload): Call finish_spills before
	restarting reload loop. Skip select_reload_regs
	if update_eliminables_and_spill returns true.

testsuite/
	PR target/60040
	* gcc.target/avr/pr60040-1.c: New.
	* gcc.target/avr/pr60040-2.c: New.

Added:
    trunk/gcc/testsuite/gcc.target/avr/pr60040-1.c
    trunk/gcc/testsuite/gcc.target/avr/pr60040-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/reload1.c
    trunk/gcc/testsuite/ChangeLog

Fixed in trunk (7.0)