Yesterday, after updating binutils to head, I hit an assembler error assembling code from checksum.h: " add,dc %%r20, %0, %0\n" It turns out that that gas was previously mis-assembling add %r1,%r2,%r3 when generating PA 2.0 code. It seems the overflow (d bit) was incorrectly being set, indicating that a 64-bit signed overflow was needed. In fact, 32-bit signed overflow is the default unless there is a "dc" completor or a 64-bit condition. A patch was applied to fix the above. However, it broke the handling of "add,dc". It also didn't fix sub. It seems that the gas bug was hiding some code generation issues in gcc. The code in the linux kernel where I noticed this is also broken. The linux change is untested. Until gcc is fixed, binutils head can't be used ;( Dave -- J. David Anglin dave.anglin@xxxxxxxxxxxxxx National Research Council of Canada (613) 990-0752 (FAX: 952-6602) diff --git a/arch/parisc/include/asm/checksum.h b/arch/parisc/include/asm/checksum.h index c84b2fc..3efcc33 100644 --- a/arch/parisc/include/asm/checksum.h +++ b/arch/parisc/include/asm/checksum.h @@ -140,7 +140,7 @@ static __inline__ __sum16 csum_ipv6_magic(const struct in6_addr *saddr, " ldd,ma 8(%1), %%r19\n" /* get 1st saddr word */ " ldd,ma 8(%2), %%r20\n" /* get 1st daddr word */ " add %8, %3, %3\n"/* add 16-bit proto + len */ -" add %%r19, %0, %0\n" +" add,* %%r19, %0, %0\n" " ldd,ma 8(%1), %%r21\n" /* 2cd saddr */ " ldd,ma 8(%2), %%r22\n" /* 2cd daddr */ " add,dc %%r20, %0, %0\n" -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
