Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[Joel Rosdahl]

On 10 December 2015 at 20:32, vkr  wrote:

> Sorry. It is not the manifest file that it tries to update. But the ".d"
> files. I think it is a real bug. [...]
>

Yes, good catch. Thanks!

Fixed like suggested in a7ab503f07e31ebeaaec34fbaa30e264308a299d
.
I also added a couple of suitable asserts in put_file_in_cache.

-- Joel
___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[Andrew Stubbs]

On 10/12/15 19:32, vkr wrote:

Sorry. It is not the manifest file that it tries to update. But the ".d"
files. I think it is a real bug.

In read-only mode, when ccache tries to run cache look up in
preprocessor mode, it tries to put_file_in_cache for .d file
unconditionally.

 if (generating_dependencies && mode == FROMCACHE_CPP_MODE) {
 put_file_in_cache(output_dep, cached_dep);
 }

This is buggy, as it should check also check we are in
read-only/read-only-direct mode, like so:

if (generating_dependencies && mode == FROMCACHE_CPP_MODE &&
(!conf->read_only && !conf->read_only_direct)) {

Or even better, if put_file_in_cache returns if we are in
read-only/read-only-direct mode, which should fix more such hidden
cases, in future.


I think you're right, and I'm sure Joel would appreciate a patch if you 
could cook one up and test it.


I'm not the maintainer, so my opinion on the placement of the fix is not 
interesting. I would follow whatever precedent has been set elsewhere, 
and only "fix" it if things are getting out of hand.


Andrew


___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[Andrew Stubbs]

On 10/12/15 03:13, vkr wrote:

Hello Andew,

The answer is "no", because it is necessary to run the preprocessor in

order to fill the cache and record the file list that direct mode lookups
use.

What are the consequences if we don't let ccache record the file list, but
just run the preprocessor and
do a cache lookup using the preprocessor mode?

WiIl I just have additional overhead when the next direct-look up runs? I
guess this is the only drawback.


If you don't record the direct-mode meta-data then you can never get a 
direct-mode cache-hit. You will rely on preprocessor mode only, for that 
input anyway.


I understood why you might want to skip the preprocessor mode lookup, or 
not bother caching compiles that will never happen again (read-only 
mode), but this seems like an optimization too far?


If a compile is usefully cachable then you should let ccache go ahead 
and do so. Yes, cache-misses are expensive, and yet the ccache overhead 
makes them even more so (my testing showed about 10% overhead, IIRC), 
but it's all for the sake of the cache-hits on subsequent runs.


I suppose the if you *know* that a direct mode check will always fail 
(modified header, perhaps), but preprocessor mode will succeed, then it 
might make sense to skip the direct mode recording so that you can skip 
the direct mode file comparison on the next run. But, in that case, you 
can just simply disable direct-mode for that file (via environment 
variables in your makefile).


If ccache had a complete set of options then you could optimize for 
every eventuality, I suppose. Maybe its worth doing?


CCACHE_DISABLE_DIRECT_LOOKUP
  - but still write meta-data
  - optimize initial builds
CCACHE_DISABLE_DIRECT_UPDATE
  - but still allow lookup
  - optimize cache-misses when direct mode would be pointless on
subsequent runs
CCACHE_DISABLE_PREPROCESSOR_LOOKUP
  - direct-mode lookup only
  - optimize cache-misses when we know processor mode will also fail
  - makes the cache read-only
CCACHE_DISABLE_PREPROCESSOR_UPDATE
  - like CCACHE_READONLY but we can still write stats

You really are getting into shaving fractions now though. At this point 
you should seriously consider linking ccache statically; run it with 
callgrind and you can see that a significant proportion of time is spent 
before "main" is even called, loading libraries and such.


Most of the rest of the time is spent doing MD4. I have some ideas how 
to optimize that (by sharing them across runs), but nothing ready to post.


Andrew

___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[vkr]

Hello Andew,
> The answer is "no", because it is necessary to run the preprocessor in
order to fill the cache and record the file list that direct mode lookups
use.

What are the consequences if we don't let ccache record the file list, but
just run the preprocessor and
do a cache lookup using the preprocessor mode?

WiIl I just have additional overhead when the next direct-look up runs? I
guess this is the only drawback.


regards,
Venkat.

On Fri, Sep 5, 2014 at 1:51 AM, Andrew Stubbs  wrote:

> On 04/09/14 08:32, vkr wrote:> I would like to know if I can avoid
> > "If there is no match, ccache falls back to running the preprocessor."
>
> The answer is "no", because it is necessary to run the preprocessor in
> order to fill the cache and record the file list that direct mode lookups
> use.
>
> It might be possible if you wanted to run the cache in read-only mode, but
> that might not be very useful, and I don't believe there's an option for
> that now.
>
> Andrew
>
___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[vkr]

Hello.
Sure, that name is an apt one and I'm glad saved a few seconds  :)

Regards,
venkrao

On Tue, Sep 16, 2014 at 12:54 AM, Joel Rosdahl j...@rosdahl.net wrote:
 Hi,

 I'm not opposed to introducing such an option. What would be a good name?
 CCACHE_READONLY_DIRECT (readonly_direct in the config file on master)?

 -- Joel


 On 10 September 2014 09:57, Andrew Stubbs a...@codesourcery.com wrote:

 On 10/09/14 08:24, vkr wrote:

 Why is that? I mean, when drop preprocessing only when an env variable is
 set,
 why is it not the right approach?


 The development sources on the master branch has all-new option and
 environment variable processing code.

 The part where you act on the option is probably in the right place, but
 it ought to be read elsewhere. I'm not sure how that worked in 3.1.9.


 Andrew
 ___
 ccache mailing list
 ccache@lists.samba.org
 https://lists.samba.org/mailman/listinfo/ccache


___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[vkr]

Hi Andrew,
 It's not the right implementation for the master branch, but might be ok in
 the release branch?

Why is that? I mean, when drop preprocessing only when an env variable is set,
why is it not the right approach?

Regards,
Venkrao

On Mon, Sep 8, 2014 at 4:32 PM, Andrew Stubbs a...@codesourcery.com wrote:
 On 08/09/14 11:37, vkr wrote:

 If the env variable TRS_CCACHE_NO_PREPROCESSOR_ON_DIRECT_LOOKUP_MISS is
 set
 then, I do not run preprocessing, but just fall back to running
 real-compiler.

 I see a lot of time saving, and I can't think of any side-effect.


 It's not the right implementation for the master branch, but might be ok in
 the release branch?

 That's up to Joel.

 Seems like a worthwhile feature, in read-only mode.

 Andrew

___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

Re: [ccache] ccacje direct mode without fall back to running the preprocessor

[vkr]

Hi Andrew
It seems to me that I can switch off preprocessing in a read-only
ccache mode, after a direct lookup fail.
How do you see the following patch?

If the env variable TRS_CCACHE_NO_PREPROCESSOR_ON_DIRECT_LOOKUP_MISS is set
then, I do not run preprocessing, but just fall back to running real-compiler.

I see a lot of time saving, and I can't think of any side-effect.

Your comments on this are very much appreciated. Thank you!.

--- ccache-3.1.9/ccache.c   2013-01-06 22:27:59.0 +0530
+++ ccache-3.1.9_patched/ccache.c   2014-09-08 15:55:49.0 +0530
@@ -2029,6 +2029,14 @@
}
  }

+
+ if (getenv(CCACHE_READONLY)) {
+   if (getenv(TRS_CCACHE_NO_PREPROCESSOR_ON_DIRECT_LOOKUP_MISS)) {
+   cc_log(TRS; Lets no run preprocessor in read-only mode, as it
goes in vain. Running real-compiler.);
+   failed();
+  }
+ }
+
  /*
   * Find the hash using the preprocessed output. Also updates
   * included_files.


On Fri, Sep 5, 2014 at 2:51 PM, vkr venkatakrishnarao...@gmail.com wrote:
 Hello
 First of all, sorry about the typo in the subject/title of my post.

 I run ccache in read-only mode itself. But, I still see that after the
 direct cache lookup fail, it falls back to running the preprocessor.
 That's exactly what I wanted to avoid.

 It seems to me that having an option to avoid preprocessing after a
 direct-cache-lookup fail is still a very valid option if the user
 knows that preprocesor will result in a failure/he purposefully wants
 to avoid it.
 I hope my justification is valid.

  88 [2014-09-05T14:45:17.445383 25208] Hostname: host.myhost.net
  89 [2014-09-05T14:45:17.445393 25208] Working directory: (null)
  90 [2014-09-05T14:45:17.445399 25208] Base directory:
 /var/fpwork/workspace_venkrao/release
  91 [2014-09-05T14:45:17.445734 25208] Source file: ../my_source.cpp
  92 [2014-09-05T14:45:17.445745 25208] Dependency file: ../my_source.d
  93 [2014-09-05T14:45:17.445752 25208] Object file: ../my_source.cpp.o
  94 [2014-09-05T14:45:17.447011 25208] Trying direct lookup
  95 [2014-09-05T14:45:17.447167 25208] Looking for object file hash in
 /disk1/ccache_cache_store/transport_release/175750/0/5/eb2b36e5e9d0bb723de833a4529cab-11859.manifest
  96 [2014-09-05T14:45:17.479757 25204] Looking for object file hash in
 /disk1/ccache_cache_store/transport_release/175750/6/6/e186baaf3c2e9e06619878dac77fe1-114310.manifest
  97 [2014-09-05T14:45:17.863524 25194] Did not find object file hash in 
 manifest
  98 [2014-09-05T14:45:17.864854 25194] Running preprocessor



 Regards,
 venkrao

 On Fri, Sep 5, 2014 at 1:51 AM, Andrew Stubbs a...@codesourcery.com wrote:
 On 04/09/14 08:32, vkr wrote: I would like to know if I can avoid
 If there is no match, ccache falls back to running the preprocessor.

 The answer is no, because it is necessary to run the preprocessor in order
 to fill the cache and record the file list that direct mode lookups use.

 It might be possible if you wanted to run the cache in read-only mode, but
 that might not be very useful, and I don't believe there's an option for
 that now.

 Andrew
___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache

[ccache] ccacje direct mode without fall back to running the preprocessor

[vkr]

Hello
The documentation about direct mode says:

The current contents of the include files are then hashed and compared
to the information in the manifest. If there is a match, ccache knows
the result of the compilation. If there is no match, ccache falls back
to running the preprocessor. The output from the preprocessor is
parsed to find the include files that were read. The paths and hash
sums of those include files are then stored in the manifest along with
information about the produced compilation result.

I would like to know if I can avoid
If there is no match, ccache falls back to running the preprocessor.

The reason is if I know that cache lookup with preprocessor output
will also result in a cache miss, there is no need to run
preprocessor,
instead, just fall back directly to running the real compiler, as this
will be faster.

Is there a way to do it?

Thanks,
-venkrao
___
ccache mailing list
ccache@lists.samba.org
https://lists.samba.org/mailman/listinfo/ccache