[Mono-devel-list] incorrect code generated by mcs
l_m at pacbell.net
Tue Nov 4 17:57:34 EST 2003
For IPF. I have it running but I recently introduced a new pointer state (regp/REGPTR) in the matcher to trigger implicit sign extension to nati when mixing int32&nati (or errors when converting a nati to int32 without a conv.i4).
I should have a 64bit patch for mini soon.
You can still work on the AMD64 codegen for mini even if the pointer arithmetic is wrong. I did just that for a long time and could bootstrap the runtime with mini w/o any problem. I only started to see SEGVs when running the unsafe tests.
zalman at peachfish.com wrote:
For which 64-bit architecture are you doing this? I'm about to start
looking at making it work on AMD64.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Mono-devel-list