Fix timing of parseFunctions.

Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the function parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None

R=jpp@chromium.org, stichnot@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=b6e9b897d9cda26ee4272c041a620195db9beaed

[Karl, 2016-03-07 21:16:03]

Description was changed from

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the funtion parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queueing costs, and 9% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None
==========

to

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the funtion parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None
==========

[Karl, 2016-03-07 21:16:45]

kschimpf@google.com changed reviewers:
+ eholk@chromium.org, jpp@chromium.org, sehr@chromium.org, stichnot@chromium.org

[John, 2016-03-07 22:40:39]

lgtm

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter.cpp
File src/IceELFObjectWriter.cpp (right):

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter....
src/IceELFObjectWriter.cpp:223: constexpr Ice::TimerStackIdT StackID =
Ice::GlobalContext::TSK_Funcs;
No need to fully qualify these.

[Jim Stichnoth, 2016-03-07 23:07:54]

Description was changed from

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the funtion parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None
==========

to

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the function parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None
==========

[Jim Stichnoth, 2016-03-07 23:20:12]

lgtm

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter.cpp
File src/IceELFObjectWriter.cpp (right):

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter....
src/IceELFObjectWriter.cpp:632: void ELFObjectWriter::setUndefinedSyms(const
ConstantList &UndefSyms) {
I think you need a TT_writeELF marker here as well.  This is based on reading
the documentation at the start of the .h file.

https://codereview.chromium.org/1775603002/diff/20001/src/IceTimerTree.def
File src/IceTimerTree.def (right):

https://codereview.chromium.org/1775603002/diff/20001/src/IceTimerTree.def#ne...
src/IceTimerTree.def:30: X(emit)                      \
What do you think about renaming this to something like emitAsm?  For a long
time I mistakenly thought that referred to ELF emission as well...

[Karl, 2016-03-08 20:27:16]

Description was changed from

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the function parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None
==========

to

==========
Fix timing of parseFunctions.

The previous implementation was charging about 24% more time that it
should to the function parser. The cause was that the time to "queue" the
parsed functions, and the time to emit the assembled code (again
including "queue" time) was not accounted for. About 15% was going to
queuing costs, and 7% to emitting the ELF file.

This CL adds timing of function translateFunctions, which captures most
of the queueing costs, and timing for each of the major ELF emission
functions (emitELF). This allows the corresponding costs to be better
bucketed, and not charged to the time it takes to parse functions in
bitcode files.

Bug=None

R=jpp@chromium.org, stichnot@chromium.org

Committed:
https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a...
==========

[Karl, 2016-03-08 20:27:17]

Committed patchset #3 (id:40001) manually as
b6e9b897d9cda26ee4272c041a620195db9beaed (presubmit successful).

[Karl, 2016-03-08 21:52:47]

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter.cpp
File src/IceELFObjectWriter.cpp (right):

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter....
src/IceELFObjectWriter.cpp:223: constexpr Ice::TimerStackIdT StackID =
Ice::GlobalContext::TSK_Funcs;
On 2016/03/07 22:40:39, John wrote:
> No need to fully qualify these.

Done.

https://codereview.chromium.org/1775603002/diff/20001/src/IceELFObjectWriter....
src/IceELFObjectWriter.cpp:632: void ELFObjectWriter::setUndefinedSyms(const
ConstantList &UndefSyms) {
On 2016/03/07 23:20:12, stichnot wrote:
> I think you need a TT_writeELF marker here as well.  This is based on reading
> the documentation at the start of the .h file.

Done.

https://codereview.chromium.org/1775603002/diff/20001/src/IceTimerTree.def
File src/IceTimerTree.def (right):

https://codereview.chromium.org/1775603002/diff/20001/src/IceTimerTree.def#ne...
src/IceTimerTree.def:30: X(emit)                      \
On 2016/03/07 23:20:12, stichnot wrote:
> What do you think about renaming this to something like emitAsm?  For a long
> time I mistakenly thought that referred to ELF emission as well...

Done.