Change Details

I recently discovered a bug on the x86 platform: The fp80 type was not handled well by x86 for constrained floating point nodes, as their regular counterparts are replaced by extending loads and truncating stores during the preprocess phase. Normally, platforms don't have this issue, as they don't typically attempt to perform such modifications during preprocessing, so strict_fp nodes will survive until they get mutated to normal nodes, which happens shortly after preprocessing on other platforms. This modification lowers these nodes at the same phase while properly utilizing the chain.

I recently discovered a bug on the x86 platform: The fp80 type was not handled well by x86 for constrained floating point nodes, as their regular counterparts are replaced by extending loads and truncating stores during the preprocess phase. Normally, platforms don't have this issue, as they don't typically attempt to perform such legalizations during instruction selection preprocessing, so strict_fp nodes will survive until they get mutated to normal nodes, which happens shortly after preprocessing on other platforms. This modification lowers these nodes at the same phase while properly utilizing the chain.

I recently discovered a bug on the x86 platform: The fp80 type was not handled well by x86 for constrained floating point nodes, as their regular counterparts are replaced by extending loads and truncating stores during the preprocess phase. Normally, platforms don't have this issue, as they don't typically attempt to perform such modificlegalizations during instruction selection preprocessing, so strict_fp nodes will survive until they get mutated to normal nodes, which happens shortly after preprocessing on other platforms. This modification lowers these nodes at the same phase while properly utilizing the chain.