diff options
| author | Aldy Hernandez <aldyh@redhat.com> | 2023-07-14 12:16:17 +0200 |
|---|---|---|
| committer | Aldy Hernandez <aldyh@redhat.com> | 2023-07-17 09:17:59 +0200 |
| commit | 56cf8b01fe1d4d4bb33a107e5d490f589d5f05bc (patch) | |
| tree | a6f850cc864497e4f67fb69a79b9ad506054e39b /gcc/value-range.h | |
| parent | 0407ae8a7732d90622a65ddf1798c9d51d450e9d (diff) | |
| download | gcc-56cf8b01fe1d4d4bb33a107e5d490f589d5f05bc.zip gcc-56cf8b01fe1d4d4bb33a107e5d490f589d5f05bc.tar.gz gcc-56cf8b01fe1d4d4bb33a107e5d490f589d5f05bc.tar.bz2 | |
Normalize irange_bitmask before union/intersect.
The bit twiddling in union/intersect for the value/mask pair must be
normalized to have the unknown bits with a value of 0 in order to make
the math simpler. Normalizing at construction slowed VRP by 1.5% so I
opted to normalize before updating the bitmask in range-ops, since it
was the only user. However, with upcoming changes there will be
multiple setters of the mask (IPA and CCP), so we need something more
general.
I played with various alternatives, and settled on normalizing before
union/intersect which were the ones needing the bits cleared. With
this patch, there's no noticeable difference in performance either in
VRP or in overall compilation.
gcc/ChangeLog:
* value-range.cc (irange_bitmask::verify_mask): Mask need not be
normalized.
* value-range.h (irange_bitmask::union_): Normalize beforehand.
(irange_bitmask::intersect): Same.
Diffstat (limited to 'gcc/value-range.h')
| -rw-r--r-- | gcc/value-range.h | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/gcc/value-range.h b/gcc/value-range.h index 0170188..d8af6fc 100644 --- a/gcc/value-range.h +++ b/gcc/value-range.h @@ -211,8 +211,12 @@ irange_bitmask::operator== (const irange_bitmask &src) const } inline bool -irange_bitmask::union_ (const irange_bitmask &src) +irange_bitmask::union_ (const irange_bitmask &orig_src) { + // Normalize mask. + irange_bitmask src (orig_src.m_value & ~orig_src.m_mask, orig_src.m_mask); + m_value &= ~m_mask; + irange_bitmask save (*this); m_mask = (m_mask | src.m_mask) | (m_value ^ src.m_value); m_value = m_value & src.m_value; @@ -222,8 +226,12 @@ irange_bitmask::union_ (const irange_bitmask &src) } inline bool -irange_bitmask::intersect (const irange_bitmask &src) +irange_bitmask::intersect (const irange_bitmask &orig_src) { + // Normalize mask. + irange_bitmask src (orig_src.m_value & ~orig_src.m_mask, orig_src.m_mask); + m_value &= ~m_mask; + irange_bitmask save (*this); // If we have two known bits that are incompatible, the resulting // bit is undefined. It is unclear whether we should set the entire |