[Rd] R-alpha/R-beta builds on Alpha platform failing with compiler error

Dirk Eddelbuettel edd at debian.org
Wed Apr 8 02:53:04 CEST 2009


On 4 April 2009 at 16:34, Dirk Eddelbuettel wrote:
| On 4 April 2009 at 23:21, Peter Dalgaard wrote:
| | Dirk Eddelbuettel wrote:
| | > Per https://buildd.debian.org/build.php?pkg=r-base, we see the Alpha platform
| | > having trouble with deriv.c :
| | > 
| | > gcc     -I. -I../../src/include -I../../src/include  -DHAVE_CONFIG_H  -mieee-with-inexact -fpic  -std=gnu99 -O3 -pipe  -g -c deriv.c -o deriv.o
| | > deriv.c: In function 'simplify':
| | > deriv.c:267: error: unrecognizable insn:
| | > (insn 2103 64 65 9 ../../src/include/Rinlinedfuns.h:86 (set (reg:DI 2 $2)
| | >         (const:DI (plus:DI (label_ref:DI 68)
| | >                 (const_int 24 [0x18])))) -1 (insn_list:REG_LABEL_OPERAND 68 (nil)))
| | > deriv.c:267: internal compiler error: in extract_insn, at recog.c:2001
| | > Please submit a full bug report,
| | > with preprocessed source if appropriate.
| | > See <file:///usr/share/doc/gcc-4.3/README.Bugs> for instructions.
| | > make[4]: *** [deriv.o] Error 1
| | > 
| | > This could of course be a bug in gcc.  Does anybody have insights into
| | > anything that may have changed here? 
| | 
| | I think it is by definition a bug in gcc...
| | 
| | It is usually rather hopeless to work around this sort of issue by 
| | source code changes. Have you tried reducing the optimization level?
| 
| No, not yet. Let me try that for the next beta. Change made.

And tested on one the Alpha machines available to Debian -- it does build
with -O2 so that's what the next upload will use.

Dirk

-- 
Three out of two people have difficulties with fractions.



More information about the R-devel mailing list