| commit | 3305a519c1b96449815a2396c77060aba0f358cf | |
| author | Alastair Bridgewater <nyef@kana.lisphacker.com> | |
| Tue, 22 Apr 2014 11:37:52 +0000 (22 07:37 -0400) | ||
| committer | Alastair Bridgewater <nyef@kana.lisphacker.com> | |
| Sat, 10 May 2014 19:12:21 +0000 (10 15:12 -0400) | ||
| tree | 9de32f68867d918fa7f08074c977cd7eca920d7d | treesnapshot (tar.gz zip) |
| parent | 90adf16eec293a5436f94e57c0a5501a938f451f | commitdiff |
runtime/arm-arch.c: Turns out there's another SIGTRAP generation method.
* Found out about this one recently. It turns out that modern
ARM Linux recognizes a specific undefined (specifically, defined
as being unlikely to be defined, similar to x86oid UD2)
instruction as a call for SIGTRAP rather than SIGILL. We'll still
want our clever conditional SWI for PSEUDO-ATOMIC, but the single
word trap should simplify things elsewhere.
* Found out about this one recently. It turns out that modern
ARM Linux recognizes a specific undefined (specifically, defined
as being unlikely to be defined, similar to x86oid UD2)
instruction as a call for SIGTRAP rather than SIGILL. We'll still
want our clever conditional SWI for PSEUDO-ATOMIC, but the single
word trap should simplify things elsewhere.
| src/runtime/arm-arch.c | diffblobblamehistory |