Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#241 Implementation of feature request #409

None
closed
None
5
2014-05-13
2014-05-13
Ben Shi
No

After applying my patch, the issue in printf_large.c is fixed, and most cases passed in regression test, except the two reported in bug #2275. (https://sourceforge.net/p/sdcc/bugs/2275/)

However these failures seems not caused by my patch.

Can anybody do a test with my patch against rev-9024 on other platforms ?

1 Attachments

Discussion

    • status: open --> closed
    • assigned_to: Philipp Klaus Krause
    • Group: -->
     
  • While replacing regFree() by regDead() here is not correct in general, the fact that it made those tests pass helped me track down the underlying issue: There was a bug in the handling of register masks in ralloc2.cc, which in turn made regFree() behave wrongly.

    I fixed the issue in revision #9025 and enabled the optimizations for bitwise and.

    Philipp

    P.S.: While I didn't apply the patch, I probably wouldn't have found the underlying issue for some time if the patch had not been there. Thanks for helping making the stm8 port more correct and more efficient.