| commit | 29058c55b6737b2db4750a0325e2207e5f97e7b6 | |
| author | Alastair Bridgewater <alastair.bridgewater@gmail.com> | |
| Thu, 4 Jan 2018 15:04:30 +0000 (4 10:04 -0500) | ||
| committer | Alastair Bridgewater <alastair.bridgewater@gmail.com> | |
| Thu, 4 Jan 2018 20:42:26 +0000 (4 15:42 -0500) | ||
| tree | b94a4a8ae67b6e019702a02d22d251b89cc42475 | treesnapshot (tar.gz zip) |
| parent | 5bf24f0e83d0f501b8a6529875366330aa13f52e | commitdiff |
ir2tran: COPY-MORE-ARG doing XEP-SETUP-SP's job is arch-specific
* COPY-MORE-ARG doing XEP-SETUP-SP is arch-specific, having
nothing to do with precise-arg-count-error. The case that this
was originally trying to protect against was taking an error trap
before the stack frame was sufficiently initialized to do a
backtrace, but that still happened on precisely the systems
supposedly being protected, and has since been fixed.
* So, rather than use a misleading tangentially-related feature
conditional, tighten up the logic generally and test for affected
architectures.
* COPY-MORE-ARG doing XEP-SETUP-SP is arch-specific, having
nothing to do with precise-arg-count-error. The case that this
was originally trying to protect against was taking an error trap
before the stack frame was sufficiently initialized to do a
backtrace, but that still happened on precisely the systems
supposedly being protected, and has since been fixed.
* So, rather than use a misleading tangentially-related feature
conditional, tighten up the logic generally and test for affected
architectures.
| src/compiler/ir2tran.lisp | diffblobblamehistory |