X87: [x64/ia32] Deal with the non-transitivity of InstructionSelector::CanCover() when folding load…

X87: [x64/ia32] Deal with the non-transitivity of InstructionSelector::CanCover() when folding loads into branches.

    port 0d22e7e46a73bebb1e92768a8443a7b7b101051b (r36482)

    original commit message:
    Sequences like:

    1: Load[kRepWord32|kTypeInt32](<address>, ...)
    2: Word32And(1, <constant>)
    3: Word32Equal(2, <another constant>)
    4: Store[(kRepWord32 : NoWriteBarrier)](<address>, <value>)
    5: Branch[None](3, ...) -> B1, B2

    where #1 and #4 refer to the same memory location, are problematic because in VisitBranch we assume that 'InstructionSelector::CanCover()' is transitive.

    What happens is that CanCover(5, 3) is true (3 is a pure op), and so are CanCover(3, 2), CanCover(2, 1), but the effect level of 5 and 3 never gets checked because 3 is a pure op. Upon VisitBranch, we

    mov [address], <value>
    test [address], <another constant>

    With this patch, it becomes:

    mov reg, [address]
    mov [address], <value>
    test reg, <another constant>

BUG=

Committed: https://crrev.com/4d9149e1befb9fa56bc498bd2467134173be85b3
Cr-Commit-Position: refs/heads/master@{#36501}

[zhengxing.li, 2016-05-25 05:07:18]

zhengxing.li@intel.com changed reviewers:
+ chunyang.dai@intel.com, epertoso@chromium.org, weiliang.lin@intel.com

[zhengxing.li, 2016-05-25 05:07:19]

PTAL, thanks!

[Weiliang, 2016-05-25 08:13:33]

lgtm

[zhengxing.li, 2016-05-25 08:34:28]

The CQ bit was checked by zhengxing.li@intel.com

[commit-bot: I haz the power, 2016-05-25 08:34:40]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/2006223004/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/2006223004/1

[commit-bot: I haz the power, 2016-05-25 08:57:30]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-05-25 09:00:04]

Description was changed from

==========
X87: [x64/ia32] Deal with the non-transitivity of
InstructionSelector::CanCover() when folding loads into branches.

    port 0d22e7e46a73bebb1e92768a8443a7b7b101051b (r36482)

    original commit message:
    Sequences like:

    1: Load[kRepWord32|kTypeInt32](<address>, ...)
    2: Word32And(1, <constant>)
    3: Word32Equal(2, <another constant>)
    4: Store[(kRepWord32 : NoWriteBarrier)](<address>, <value>)
    5: Branch[None](3, ...) -> B1, B2

    where #1 and #4 refer to the same memory location, are problematic because
in VisitBranch we assume that 'InstructionSelector::CanCover()' is transitive.

    What happens is that CanCover(5, 3) is true (3 is a pure op), and so are
CanCover(3, 2), CanCover(2, 1), but the effect level of 5 and 3 never gets
checked because 3 is a pure op. Upon VisitBranch, we

    mov [address], <value>
    test [address], <another constant>

    With this patch, it becomes:

    mov reg, [address]
    mov [address], <value>
    test reg, <another constant>

BUG=
==========

to

==========
X87: [x64/ia32] Deal with the non-transitivity of
InstructionSelector::CanCover() when folding loads into branches.

    port 0d22e7e46a73bebb1e92768a8443a7b7b101051b (r36482)

    original commit message:
    Sequences like:

    1: Load[kRepWord32|kTypeInt32](<address>, ...)
    2: Word32And(1, <constant>)
    3: Word32Equal(2, <another constant>)
    4: Store[(kRepWord32 : NoWriteBarrier)](<address>, <value>)
    5: Branch[None](3, ...) -> B1, B2

    where #1 and #4 refer to the same memory location, are problematic because
in VisitBranch we assume that 'InstructionSelector::CanCover()' is transitive.

    What happens is that CanCover(5, 3) is true (3 is a pure op), and so are
CanCover(3, 2), CanCover(2, 1), but the effect level of 5 and 3 never gets
checked because 3 is a pure op. Upon VisitBranch, we

    mov [address], <value>
    test [address], <another constant>

    With this patch, it becomes:

    mov reg, [address]
    mov [address], <value>
    test reg, <another constant>

BUG=

Committed: https://crrev.com/4d9149e1befb9fa56bc498bd2467134173be85b3
Cr-Commit-Position: refs/heads/master@{#36501}
==========

[commit-bot: I haz the power, 2016-05-25 09:00:05]

Patchset 1 (id:??) landed as
https://crrev.com/4d9149e1befb9fa56bc498bd2467134173be85b3
Cr-Commit-Position: refs/heads/master@{#36501}