Re: [kbuild-all] Re: drivers/mtd/tests/subpagetest.c:426:1: error: could not split insn

From: Rong Chen
Date: Wed Dec 16 2020 - 02:21:41 EST




On 12/15/20 11:40 PM, Willy Tarreau wrote:
Hi,

On Tue, Dec 15, 2020 at 11:05:28PM +0800, kernel test robot wrote:
Hi Willy,

FYI, the error/warning still remains.

tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
head: 148842c98a24e508aecb929718818fbf4c2a6ff3
commit: 3744741adab6d9195551ce30e65e726c7a408421 random32: add noise from network and scheduling activity
(...)

not sure why I'm assigned this, but the root cause is a compiler bug:
drivers/mtd/tests/subpagetest.c: In function 'mtd_subpagetest_init':
drivers/mtd/tests/subpagetest.c:426:1: error: could not split insn
426 | }
| ^
(insn:TI 453 2652 455 (set (reg/v:SI 3 a3 [orig:304 a ] [304])
(xor:SI (reg:SI 1 a1 [orig:717 net_rand_noise ] [717])
(const:SI (plus:SI (symbol_ref:SI ("*.LANCHOR0") [flags 0x182])
(const_int 12 [0xc]))))) "include/linux/prandom.h":66:4 152 {cskyv2_xorsi3}
(expr_list:REG_DEAD (reg:SI 1 a1 [orig:717 net_rand_noise ] [717])
(nil)))
during RTL pass: final
drivers/mtd/tests/subpagetest.c:426:1: internal compiler error: in final_scan_insn_1, at final.c:3074
^^^^^^^^^^^^^^^^^^^^^^^

0x510da0 _fatal_insn(char const*, rtx_def const*, char const*, int, char const*)
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/rtl-error.c:108
0x503d22 final_scan_insn_1
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/final.c:3074
0x73f8bf final_scan_insn(rtx_insn*, _IO_FILE*, int, int, int*)
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/final.c:3153
0x73fbac final_1
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/final.c:2021
0x740618 rest_of_handle_final
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/final.c:4659
0x740618 execute
/tmp/build-crosstools-xh-9.3.0-2.34/gcc/gcc-9.3.0/gcc/final.c:4737
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
^^^^^^^^^^^^^^^^^^^^^^^^^

That's totally out of my scope. I suspect it might have broken a bisect
operation.

Hi Willy,

Thanks for the feedback, I have created a issue in GCC Bugzilla:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98310

Best Regards,
Rong Chen


Regards,
Willy
_______________________________________________
kbuild-all mailing list -- kbuild-all@xxxxxxxxxxxx
To unsubscribe send an email to kbuild-all-leave@xxxxxxxxxxxx