Please take a look the updated patch. It addresses the issue of using
dlclose before dump, and potential races (between a thread closing a
library and the dumper call).
David
On Sun, Jul 20, 2014 at 11:12 PM, Nathan Sidwell nat...@acm.org wrote:
On 07/20/14 21:38, Xinliang David Li wrote
with shared libs.
David
On Sun, Jul 20, 2014 at 12:42 PM, Nathan Sidwell nat...@acm.org wrote:
On 07/18/14 22:41, Xinliang David Li wrote:
Hi, the following patch implements a new dumper interface to allow
dumping of profile data for all instrumented shared libraries.
For good reasons, existing
Any other concern of the patch? I don't really like the name of the
parameter myself. Do you have better suggestions?
thanks,
David
On Thu, Jul 17, 2014 at 10:17 AM, Xinliang David Li davi...@google.com wrote:
I see why you do not like first_global_object_name because changing it
would
===
--- libgcc/ChangeLog(revision 212682)
+++ libgcc/ChangeLog(working copy)
@@ -1,3 +1,9 @@
+2014-07-18 Xinliang David Li davi...@google.com
+
+ * libgcov-driver.c: Force __gcov_dump to be exported
+ * libgcov-interface.c
On Wed, Jul 16, 2014 at 4:42 PM, Jan Hubicka hubi...@ucw.cz wrote:
Instrumentation based FDO is designed to work when the source files
that are used to generate the instr binary match exactly with the
sources in profile-use compile. It is known historically that using
stale profile (due to
I see why you do not like first_global_object_name because changing it would
cause
all functions from that unit to drop the profiles. Perhaps we can combine
function name
and compilation unit (gcov file) name?
that is a good idea -- it will also solve the LTO problem you mentioned above.
blocks. */
DEFPARAM (PARAM_SLP_MAX_INSNS_IN_BB,
slp-max-insns-in-bb,
Index: ChangeLog
===
--- ChangeLog (revision 212682)
+++ ChangeLog (working copy)
@@ -1,3 +1,10 @@
+2014-07-16 Xinliang David Li davi...@google.com
ok. This is now consistent with how cgraph targets are handled.
(Changes like this may shake up more bugs due to bad assumptions. More
extensive testing after this is expected).
David
On Tue, Jul 15, 2014 at 3:24 AM, Teresa Johnson tejohn...@google.com wrote:
This patch will rewrite references
On Mon, Jul 14, 2014 at 2:37 PM, Andrew Pinski pins...@gmail.com wrote:
On Mon, Jul 14, 2014 at 1:40 PM, Dehao Chen de...@google.com wrote:
This patch replaces getline with fgets so that gcc builts fine in darwin.
Why not add getline to libiberty instead? This patch causes a huge stack
I wonder why. The struct definition for gcov_fn_info has not changed
in this patch.
David
On Fri, Jul 11, 2014 at 1:07 AM, Richard Biener
richard.guent...@gmail.com wrote:
On Mon, May 5, 2014 at 7:17 PM, Rong Xu x...@google.com wrote:
Here is the updated patch. The difference with patch set 3
What is the macro to test POSIX IO on host? The guard uses
TARGET_POSIX_IO which is not right.
David
On Fri, Jul 11, 2014 at 1:12 AM, Christophe Lyon
christophe.l...@linaro.org wrote:
On 11 July 2014 10:07, Richard Biener richard.guent...@gmail.com wrote:
On Mon, May 5, 2014 at 7:17 PM, Rong
it as it is (if not doing memcpy for the whole
array).
David
On Fri, Jul 11, 2014 at 8:44 AM, Jakub Jelinek ja...@redhat.com wrote:
On Fri, Jul 11, 2014 at 08:42:27AM -0700, Xinliang David Li wrote:
I wonder why. The struct definition for gcov_fn_info has not changed
in this patch.
Perhaps it has been used
Cong, can you ping this patch again? There does not seem to be
pending comments left.
David
On Tue, Dec 17, 2013 at 10:05 AM, Cong Hou co...@google.com wrote:
Ping?
thanks,
Cong
On Mon, Dec 2, 2013 at 5:02 PM, Cong Hou co...@google.com wrote:
Any comment on this patch?
thanks,
Cong
Sri, can you provide examples to show why putting thunks into the same
section as the target function when function reorder is on can be bad
?
Thanks,
David
On Thu, Jun 26, 2014 at 10:29 AM, Sriraman Tallam tmsri...@google.com wrote:
Hi Honza,
Could you review this patch when you find
Should the interface be something like:
int __gcov_profiling_for_test_coverage(void)?
David
On Wed, Jul 2, 2014 at 12:55 PM, Teresa Johnson tejohn...@google.com wrote:
The following patch adds support for a new libgcov interface,
__gcov_profiling_enabled, that can be used by applications to
David Li davi...@google.com wrote:
Should the interface be something like:
int __gcov_profiling_for_test_coverage(void)?
I was equating the term profiling with -fprofile-generate, as
opposed to -ftest-coverage instrumentation. But I can change it to
this if you think that is clearer.
Teresa
-generate). */
+
+unsigned int __gcov_profiling_for_test_coverage (void)
+{
+ return __gcov_test_coverage;
+}
+
#endif /* L_gcov_dump */
#ifdef L_gcov_sampling
On Wed, Jul 2, 2014 at 1:25 PM, Xinliang David Li davi...@google.com wrote:
The reason is that test coverage is using
Ok, with minor fix to the comments:
1) Before possibly cloning -- Before possibly creating a new
2) It is not clear from the comments that the remaining references are
from call statements in ipa cp clones that have not been fixed up.
Possibly add a 3 line example to show the call statements
when appropriate.
* common.opt (fcheck-branch-annotation)
(fcheck-branch-annotation-threshold=): Add an extra GCC
option to turn
on report
On Fri, Jun 27, 2014 at 3:20 PM, Xinliang David Li
davi...@google.com wrote:
Hi Yi,
1) please add comments before new functions
Hi Yi,
1) please add comments before new functions as documentation -- follow
the coding style guideline
2) missing documenation on the new flags (pointed out by Gerald)
3) Please refactor the check code in afdo_calculate_branch_prob into a
helper function
4) the change log is not needed for
The change makes gcov_info a variable length array, which is not ideal.
Better just add one more field (instead of two):
struct gcov_info {
...
char ** build_info;
};
For regular case, it is null, for case where the build info is
available, make it point to a string array (with an null end
On Fri, May 23, 2014 at 1:59 PM, Teresa Johnson tejohn...@google.com wrote:
Done. Passes manual testing, rerunning regression testing. New patch attached.
Thanks,
Teresa
On Fri, May 23, 2014 at 11:35 AM, Xinliang David Li davi...@google.com
wrote:
The change makes gcov_info a variable length
On Fri, May 23, 2014 at 2:50 PM, Teresa Johnson tejohn...@google.com wrote:
On Fri, May 23, 2014 at 2:28 PM, Xinliang David Li davi...@google.com wrote:
for (i = 0; i num_strings; i++)
+{
+ if (strcmp (build_info_strings[i], gi_ptr-build_info[i]))
Add also check
ok.
David
On Fri, May 23, 2014 at 3:03 PM, Dehao Chen de...@google.com wrote:
This patch fixes LIPO ICE that an unresolved node escaped after lipo fixup.
testing on going. OK for google-4_9?
Thanks,
Dehao
Index: gcc/ipa.c
ok
On Fri, May 23, 2014 at 3:14 PM, Teresa Johnson tejohn...@google.com wrote:
On May 23, 2014 2:56 PM, Xinliang David Li davi...@google.com wrote:
On Fri, May 23, 2014 at 2:50 PM, Teresa Johnson tejohn...@google.com
wrote:
On Fri, May 23, 2014 at 2:28 PM, Xinliang David Li davi
===
--- ChangeLog (revision 210881)
+++ ChangeLog (working copy)
@@ -1,3 +1,9 @@
+2014-05-23 Xinliang David Li davi...@google.com
+
+ * tree-pretty-print.c (dump_function_header): Print cgraph uid.
+ * passes.c (pass_init_dump_file
Also missing documentation in invoke.texi? Other than that, ok.
David
On Thu, May 22, 2014 at 5:00 PM, Paul Pluzhnikov ppluzhni...@google.com wrote:
On Thu, May 22, 2014 at 4:36 PM, Sriraman Tallam tmsri...@google.com wrote:
This patch is pending review for trunk. Please see if this is ok to
Since it is in trunk and gcc-4_9-branch, there is no need to put into
google branches -- will get it automatically after next merge.
David
On Wed, May 21, 2014 at 8:13 PM, Carrot Wei car...@google.com wrote:
Committed to trunk, 4.9, and waiting for the release of 4.8.3.
OK for google/main and
Why duplicating the merger functions in dyn-ipa.c? Should those in
libgcov-merge.c be reused?
The refactoring of gcov_exit_write_gcda should probably be done in a
separate patch -- preferably submitted to trunk too.
David
On Mon, May 19, 2014 at 10:08 PM, Teresa Johnson tejohn...@google.com
On Tue, May 20, 2014 at 6:32 AM, Teresa Johnson tejohn...@google.com wrote:
On Mon, May 19, 2014 at 11:51 PM, Xinliang David Li davi...@google.com
wrote:
Why duplicating the merger functions in dyn-ipa.c? Should those in
libgcov-merge.c be reused?
The merger functions in libgcov-merge.c use
On Tue, May 20, 2014 at 4:32 AM, Richard Biener
richard.guent...@gmail.com wrote:
On Mon, May 19, 2014 at 5:24 PM, Xinliang David Li davi...@google.com wrote:
Sorry about it. Here is the patch. There is one remaining case where
cgraph_dump_file and dump_enable_p are checked separately
, 2014 at 11:19 PM, Xinliang David Li davi...@google.com
wrote:
Modified the patch according to yours and Richard's feedback. PTAL.
ENOPATCH.
Btw, I don't see any issue with leaking node order to opt-report.
Richard.
thanks,
David
On Fri, May 16, 2014 at 9:03 AM, Jan Hubicka hubi
There is no test regression. Ok with this patch?
David
On Fri, May 16, 2014 at 2:19 PM, Xinliang David Li davi...@google.com wrote:
Modified the patch according to yours and Richard's feedback. PTAL.
thanks,
David
On Fri, May 16, 2014 at 9:03 AM, Jan Hubicka hubi...@ucw.cz wrote:
Hi
Ok to check in the test?
David
On Fri, May 16, 2014 at 4:58 PM, Xinliang David Li davi...@google.com wrote:
This test makes sure compiler does not wrongly devirtualize virtual
calls into __cxa_pure_virtual or __buitlin_unreachable.
Ok to checkin?
David
On Fri, May 16, 2014 at 4:00 AM, Richard Biener
richard.guent...@gmail.com wrote:
On Fri, May 16, 2014 at 1:54 AM, Xinliang David Li davi...@google.com wrote:
Hi, debugging runtime bugs due to devirtualization can be hard for
very large C++ programs with complicated class hierarchy. This patch
)
@@ -1,3 +1,18 @@
+2014-05-15 Xinliang David Li davi...@google.com
+
+ * cgraphunit.c (walk_polymorphic_call_targets): Add
+ dbgcnt and fopt-info support.
+ 2014-05-15 Xinliang David Li davi...@google.com
+
+ * cgraphunit.c (walk_polymorphic_call_targets): Add
On Fri, May 16, 2014 at 9:51 AM, Jan Hubicka hubi...@ucw.cz wrote:
I have been chasing a runtime failure of a very large test built with
gcc-4_9. The bad code either calls a pure function or turn a virtual
call into __builtin_unreachable (incomplete target set). The indirect
info shows the
===
--- ChangeLog (revision 210479)
+++ ChangeLog (working copy)
@@ -1,3 +1,18 @@
+2014-05-15 Xinliang David Li davi...@google.com
+
+ * cgraphunit.c (walk_polymorphic_call_targets): Add
+ dbgcnt and fopt-info support.
+ 2014-05-15 Xinliang David Li davi
/ChangeLog (working copy)
@@ -1,3 +1,7 @@
+2014-05-16 Xinliang David Li davi...@google.com
+
+ * g++.dg/ipa/devirt-33.C: New testcase.
+
2014-05-15 Martin Jambor mjam...@suse.cz
PR ipa/61085
Index: testsuite/g++.dg/ipa/devirt-33.C
for trunk after build and test?
thanks,
David
Index: ChangeLog
===
--- ChangeLog (revision 210479)
+++ ChangeLog (working copy)
@@ -1,3 +1,18 @@
+2014-05-15 Xinliang David Li davi...@google.com
+
+ * cgraphunit.c
On Tue, May 13, 2014 at 1:39 AM, Richard Biener
richard.guent...@gmail.com wrote:
On Fri, May 9, 2014 at 5:54 PM, Teresa Johnson tejohn...@google.com wrote:
I discovered that the support for the documented -fdump-* options
optimized, missed, note and optall was missing. Added that and
fixed a
I have concerns with the proposed this patch:
1) not sharing cd_root may lead to difficulties in later predication
simplication
2) the change to check post-dom may also lead to incomplete predicate chain.
I think the right fix to the problem is to realize that BBs with the
following conditions
I think the right fix to the problem is to realize that BBs with the
following conditions y_8 !=0, p.0_10 !=0, and x_5 !=0 are actually
control equivalent. This fact allows simplifying the USE predicates
from (y_8 !=0 OR p.0_10 !=0 OR x_5 !=0) into just p.0_10 !=0 which is
the same as the
Ok. For non-autofdo path, I think a warning (even an error) is needed
in case of mismatch.
David
On Mon, May 12, 2014 at 4:35 PM, Rong Xu x...@google.com wrote:
Hi,
This patch adds an extra check of the incompatibility of use and saved
command line for the primary module. If the check files,
===
--- gcc/builtins.c(revision 210019)
+++ gcc/builtins.c(working copy)
@@ -87,6 +87,9 @@ static rtx result_vector (int, rtx);
#endif
static void expand_builtin_update_setjmp_buf (rtx);
static void
ok.
David
On Thu, May 8, 2014 at 1:33 PM, Rong Xu x...@google.com wrote:
Hi,
This patch fixed lipo ICE triggered by an out-of-bound access.
This is google specific patch and tested with bootstrap and the
program exposed the issue.
Thanks,
-Rong
Have you announced the autofdo profile tool to gcc list?
David
On Wed, May 7, 2014 at 2:24 PM, Dehao Chen de...@google.com wrote:
Hi,
I'm planning to port the AutoFDO patch upstream. Attached is the
prepared patch. You can also find the patch in
http://codereview.appspot.com/99010043
I've
ok. Also needed in google/gcc-4_9 branch.
David
On Mon, May 5, 2014 at 8:40 PM, Dehao Chen de...@google.com wrote:
This patch handles TYPE_PACK_EXPANSION in lipo_cmp_type.
testing on going. OK for google-4_8?
Thanks,
Dehao
Index: gcc/l-ipo.c
On Tue, Apr 22, 2014 at 1:17 PM, Jan Hubicka hubi...@ucw.cz wrote:
This looks fine. LIPO has similar change too. Other directions worth
looking into:
1) To model icache effect better, weighted callee size need to be
used with profile. The weight for BB may look like: min(1,
Bin, when will the patch for the generic pass be available for review?
David
On Wed, Apr 9, 2014 at 7:27 PM, Bin.Cheng amker.ch...@gmail.com wrote:
On Thu, Apr 10, 2014 at 8:18 AM, Wei Mi w...@google.com wrote:
Hi,
For the testcase 1.c
#include emmintrin.h
double a[1000];
__m128d
Do you witness similar problems with LTO +FDO?
My concern is it can be tricky to get the register pressure estimate
right. The register pressure problem is created by downstream
components (code motions etc) but only exposed by the inliner. If you
want to get it 'right' (i.e., not exposing the
On Fri, Apr 18, 2014 at 10:26 AM, Jan Hubicka hubi...@ucw.cz wrote:
Hello,
Honza,
Seeing your recent patches relating to inliner heuristics for LTO,
I thought I should mention some related work I'm doing.
By way of introduction, I've recently joined the IBM LTC's PPC
Toolchain team,
-- any
callsite with a net saving will be considered.
David
Honza
Aaron
On Fri, 2014-04-18 at 10:36 -0700, Xinliang David Li wrote:
Do you witness similar problems with LTO +FDO?
My concern is it can be tricky to get the register pressure estimate
right. The register pressure
On Fri, Apr 18, 2014 at 12:51 PM, Jan Hubicka hubi...@ucw.cz wrote:
What I've observed on power is that LTO alone reduces performance and
LTO+FDO is not significantly different than FDO alone.
I agree that an exact estimate of the register pressure would be a
difficult problem. I'm hoping
On Fri, Apr 18, 2014 at 2:16 PM, Jan Hubicka hubi...@ucw.cz wrote:
On Fri, Apr 18, 2014 at 12:27 PM, Jan Hubicka hubi...@ucw.cz wrote:
What I've observed on power is that LTO alone reduces performance and
LTO+FDO is not significantly different than FDO alone.
On SPEC2k6?
This is quite
This looks fine. LIPO has similar change too. Other directions worth
looking into:
1) To model icache effect better, weighted callee size need to be
used with profile. The weight for BB may look like: min(1,
FREQ(BB)/FREQ(ENTRY)).
2) When function splitting is turned on, are any inline
Looks good to me.
David
On Thu, Apr 10, 2014 at 2:04 PM, Teresa Johnson tejohn...@google.com wrote:
This patch fixes an issue where self-recursive calls were missed
during ipa inlining in LIPO compiles, since the resolved nodes were
not checked. When a self-recursive call was inlined
ok (after fixing the format -- a function name starts a new line in
function def).
David
On Mon, Apr 7, 2014 at 12:49 PM, Dehao Chen de...@google.com wrote:
This patch calls add_fake_edge for the AutoFDO+LIPO path.
Bootstrapped and passed regression test and performance test.
OK for
looks fine.
David
On Thu, Apr 3, 2014 at 10:56 AM, Dehao Chen de...@google.com wrote:
This patch updates SSA after VPT transformation. This is needed
because compute_inline_parameters will ICE without updated SSA.
Testing on-going.
OK for google-4_8?
Thanks,
Dehao
Index:
ok.
On Thu, Mar 27, 2014 at 9:02 AM, Dehao Chen de...@google.com wrote:
On Wed, Mar 26, 2014 at 4:05 PM, Xinliang David Li davi...@google.com wrote:
is cgraph_init_gid_map called after linking?
Oh, forgot that part. It's interesting that the test can pass without
another cgraph_init_gid_map
is cgraph_init_gid_map called after linking?
David
On Wed, Mar 26, 2014 at 3:54 PM, Dehao Chen de...@google.com wrote:
Patch updated, passed performance tests.
Dehao
On Tue, Mar 25, 2014 at 4:03 PM, Xinliang David Li davi...@google.com wrote:
Add comment to the new function. init_node_map
I think the right way to fix this is to wrap the call to early_inliner
and check the TODO flags. See execute_function_todo:
if (flags TODO_cleanup_cfg)
{
cleanup_tree_cfg ();
if (!(flags TODO_update_ssa_any) need_ssa_update_p (cfun))
flags |= TODO_update_ssa;
}
(promoted_stmts);
compute_function_frequency ();
update_ssa (TODO_update_ssa);
Is this update still needed?
David
On Thu, Mar 20, 2014 at 11:36 AM, Xinliang David Li davi...@google.com
wrote:
I think the right way to fix this is to wrap the call to early_inliner
and check the TODO
The patch is ok.
David
On Thu, Mar 20, 2014 at 1:10 PM, Dehao Chen de...@google.com wrote:
On Thu, Mar 20, 2014 at 1:02 PM, Xinliang David Li davi...@google.com wrote:
On Thu, Mar 20, 2014 at 12:40 PM, Dehao Chen de...@google.com wrote:
Patch updated to add a wrapper early_inline function
Why is it not enough to emit warnings during build time when source
code changes signifcantly?
David
On Tue, Mar 11, 2014 at 4:27 PM, Dehao Chen de...@google.com wrote:
During AutoFDO annotation, we want to record the annotation stats into
an elf section, so that we can calculate how much
Dehao explained that the data needs to merged during link time to give
meaningful diagnostics.
Ok for google branch.
David
On Wed, Mar 12, 2014 at 3:55 PM, Xinliang David Li davi...@google.com wrote:
Why is it not enough to emit warnings during build time when source
code changes
Ok.
David
On Mon, Mar 10, 2014 at 1:28 PM, Teresa Johnson tejohn...@google.com wrote:
This patch adds dummy references to libgcov for the symbols accessed
via weak references from application code to ensure they are resolved
at link time. Passes regression tests. Ok for google-4_8?
Thanks,
On Wed, Feb 26, 2014 at 3:23 PM, Dehao Chen de...@google.com wrote:
This patch fixes the bug of not calling compute_inline_parameters
before early_inliner, which would lead to ICE.
Testing on going, OK for google-4_8 if test passes?
Thanks,
Dehao
Index: gcc/auto-profile.c
::afdo_annotate_cfg (promoted_stmts);
compute_function_frequency ();
Dehao
On Wed, Feb 26, 2014 at 3:25 PM, Xinliang David Li davi...@google.com wrote:
On Wed, Feb 26, 2014 at 3:23 PM, Dehao Chen de...@google.com wrote:
This patch fixes the bug of not calling compute_inline_parameters
before
thanks for the fix!
David
On Fri, Feb 21, 2014 at 12:21 AM, Jakub Jelinek ja...@redhat.com wrote:
Hi!
As discussed in the PR, on larger functions we can end up with
over 3 million of compute_control_dep_chain nested calls from
a single compute_control_dep_chain call, on that testcase all
Ok. I expect this also submitted to trunk later.
David
On Fri, Feb 21, 2014 at 2:08 PM, Rong Xu x...@google.com wrote:
Hi,
For builtins without nothrow attributes, we currently split bb by adding
fake edge to func_exit in instrumenting profile counters. While it's safe,
The resulted
On Thu, Feb 13, 2014 at 9:48 AM, Mike Stump mikest...@comcast.net wrote:
On Feb 13, 2014, at 8:41 AM, Teresa Johnson tejohn...@google.com wrote:
On Wed, Feb 12, 2014 at 2:03 PM, Xinliang David Li davi...@google.com
wrote:
[ extra lines deleted ]
Should non comdat function be skipped?
We
On Wed, Feb 12, 2014 at 8:02 AM, Teresa Johnson tejohn...@google.com wrote:
On Wed, Feb 12, 2014 at 6:45 AM, Teresa Johnson tejohn...@google.com wrote:
On Tue, Feb 11, 2014 at 6:13 PM, Xinliang David Li davi...@google.com
wrote:
On Tue, Feb 11, 2014 at 5:36 PM, Teresa Johnson tejohn
On Mon, Feb 10, 2014 at 11:48 PM, Wei Mi w...@google.com wrote:
Here is the updated patch, which follow UD chain to determine whether
iv.base is defined by __gcovx.xxx[] var. It is a lot simpler than
adding a tree bit.
regression test and previously failed benchmark in piii mode is ok.
Other
ok for google branch for now. Please resubmit to trunk and backport
the trunk version if needed.
thanks,
David
On Tue, Feb 11, 2014 at 10:29 AM, Wei Mi w...@google.com wrote:
+/* Return true if the use is defined by a gcov counter var.
+ It is used to check if an iv candidate is generated
Is it better to add some logic in counts_to_freq to determine if the
profile count needs to be dropped completely to force profile
estimation?
David
On Mon, Feb 10, 2014 at 2:12 PM, Teresa Johnson tejohn...@google.com wrote:
This patch attempts to address the lost profile issue for COMDATs in
ok.
David
On Tue, Feb 11, 2014 at 4:50 PM, Cong Hou co...@google.com wrote:
This small patch lets GCC emit a single unaligned load/store
instruction for m_GENERIC i386 CPUs.
Bootstrapped and passed regression test.
OK for Google branch?
thanks,
Cong
Index: gcc/config/i386/i386.c
On Tue, Feb 11, 2014 at 5:04 PM, Teresa Johnson tejohn...@google.com wrote:
On Tue, Feb 11, 2014 at 2:56 PM, Xinliang David Li davi...@google.com wrote:
Is it better to add some logic in counts_to_freq to determine if the
profile count needs to be dropped completely to force profile
estimation
ok.
David
On Tue, Feb 11, 2014 at 5:30 PM, Cong Hou co...@google.com wrote:
With this patch x_flag_complex_method won't be set to 2 for C++ so
that multiply/divide between std::complex objects won't be replaced by
expensive builtin function calls.
Bootstrapped and passed regression test.
On Tue, Feb 11, 2014 at 5:36 PM, Teresa Johnson tejohn...@google.com wrote:
On Tue, Feb 11, 2014 at 5:16 PM, Xinliang David Li davi...@google.com wrote:
Why is call graph needed to determine whether to drop the profile?
Because we detect this situation by looking for cases where the call
edge
I don't think you should add a new tree bit just for this. A simple
UD check can determine prephitmp_23, etc are abnormal ssa names:
# prephitmp_23 = PHI PROF_edge_counter_10(5), pretmp_1(3)
PROF_edge_counter_10 = prephitmp_23 + 1;
__gcov0.foo[0] = PROF_edge_counter_10;
Note you may need
On Fri, Feb 7, 2014 at 1:22 AM, Richard Biener
richard.guent...@gmail.com wrote:
On Thu, Feb 6, 2014 at 10:30 PM, Xinliang David Li davi...@google.com wrote:
Hi the following patch removes the 'state' print for
-ftree-tree-vectorize option which does not make sense anymore. Ok for
trunk?
Hmm
)
@@ -1,3 +1,7 @@
+2014-02-06 Xinliang David Li davi...@google.com
+
+ * opts.c (print_filtered_help): Fix help message bug.
+
2014-02-06 Kyrylo Tkachov kyrylo.tkac...@arm.com
* config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
Index: opts.c
ok.
David
On Tue, Feb 4, 2014 at 11:28 AM, Rong Xu x...@google.com wrote:
Hi,
The attached patch fixes the duplicated definition error in
gcov-tool.c in profiledbootstrap.
Google branch only and tested with profiledbootstrap.
Ok for checking in?
-Rong
On Wed, Jan 29, 2014 at 4:24 PM, Rong Xu x...@google.com wrote:
Hi,
The attached patch adds some new features to gcov-tool. It aims to
rewrite LIPO profile for reuse, debug or performance tuning purpose.
-l, --modu_list file Only use the modules in this file
I think in verbose
Ok.
For future cleanup, It will be better to define a new helper function
is_profile_missing_p (cfun) -- which checks entry count for
instrumentation based FDO, and something else for autofdo.
David
On Fri, Jan 24, 2014 at 9:28 PM, Dehao Chen de...@google.com wrote:
This patch fixes
patterns with one
exception, pr53265.c. Loop unrolling exposed the array out of bounds
access which is now caught.
Ok to commit?
Thanks
Sri
On Tue, Jan 21, 2014 at 4:51 PM, Xinliang David Li davi...@google.com wrote:
ok.
David
On Tue, Jan 21, 2014 at 4:46 PM, Sriraman Tallam tmsri
for AVG_LOOP_ITER.
Both patches are attached.
Thanks,
Wei.
On Wed, Jan 22, 2014 at 4:42 PM, Xinliang David Li davi...@google.com wrote:
On Wed, Jan 22, 2014 at 2:23 PM, Wei Mi w...@google.com wrote:
This patch handles the mem access builtins in ivopt. The original
problem described here
ok.
David
On Thu, Jan 23, 2014 at 4:21 PM, Rong Xu x...@google.com wrote:
Hi,
This patch is for google/gcc-4_8 branch. It fixes a regression in
earlier libgcov refactoring.
Thanks,
-Rong
2014-01-23 Rong Xu x...@google.com
* libgcov-driver.c (__gcov_get_profile_prefix):
Is it possible to add a test case?
David
On Wed, Jan 22, 2014 at 6:08 PM, Dehao Chen de...@google.com wrote:
During floating point comparison, compiler inserts conditional jumps
to check if the operand is NAN. These type of checks are normally
false. However, compiler sets the probability the
not necessary. We only need the first part (tree-inline.c).
Thanks,
Dehao
On Fri, Jan 17, 2014 at 3:12 PM, Xinliang David Li davi...@google.com wrote:
Can callgraph node count be fixed up properly instead of doing
individual fixups like this?
David
On Fri, Jan 17, 2014 at 2:38 PM, Dehao
On Wed, Jan 22, 2014 at 2:23 PM, Wei Mi w...@google.com wrote:
This patch handles the mem access builtins in ivopt. The original
problem described here:
http://gcc.gnu.org/ml/gcc-patches/2013-11/msg02648.html
Bootstrapped and passed regression test. Performance test ok for
plain, fdo and
I think it might be better to introduce a new parameter for max peel
insn at O2 (e.g, call it MAX_O2_COMPLETELY_PEEL_INSN or
MAX_DEFAULT_...), and use the same logic in your patch to override the
MAX_COMPLETELY_PEELED_INSN parameter at O2).
By so doing, we don't need to have a hard coded factor
ok.
David
On Tue, Jan 21, 2014 at 4:46 PM, Sriraman Tallam tmsri...@google.com wrote:
On Tue, Jan 21, 2014 at 2:49 PM, Xinliang David Li davi...@google.com wrote:
I think it might be better to introduce a new parameter for max peel
insn at O2 (e.g, call it MAX_O2_COMPLETELY_PEEL_INSN
Ok.
David
On Fri, Jan 17, 2014 at 9:08 AM, Dehao Chen de...@google.com wrote:
If a loop is cunrolled/vectorized, the AutoFDO computed trip count
will be very small. This patch disallows overwritting of precomputed
loop bound in AutoFDO mode.
Bootstrapped and passed regression test.
For LIPO parameters, you can do this
1) isolate all LIPO specific parameters into lipo_params.def, and
include it from params.def.
2) include lipo_params.def (after proper definition of DEFPARAM) in
the profile tool source dir.
By so doing, the compiler setting and profile tool will be in sync.
and let logic in dyn-ipa.c takes
care of the rest.
Please find the update patch which reads the default values from params.def.
-Rong
On Fri, Jan 17, 2014 at 12:05 PM, Xinliang David Li davi...@google.com
wrote:
For LIPO parameters, you can do this
1) isolate all LIPO specific parameters
Can callgraph node count be fixed up properly instead of doing
individual fixups like this?
David
On Fri, Jan 17, 2014 at 2:38 PM, Dehao Chen de...@google.com wrote:
In AutoFDO, sometime edge count might be propagated to be too large
due to bad debug info. In this cases, we need to make sure
ok.
David
On Wed, Jan 15, 2014 at 9:10 AM, Teresa Johnson tejohn...@google.com wrote:
Regression tested, ok for google/4_8?
Thanks,
Teresa
2014-01-15 Teresa Johnson tejohn...@google.com
* loop-unroll.c (code_size_limit_factor): Handle int overflow.
Index: loop-unroll.c
In libgcov-driver.c,
1) there are couple of places with trailing white spaces (e.g, in
gcov_sort_n_vals body), please remove
2) gcov_exit_write_gcda in trunk takes eof_pos as an arg, and check it
before writing the header. I think this is more correct than in your
patch
3) It would be better to
101 - 200 of 970 matches
Mail list logo