ton Jones
Date: Wed Sep 7 09:33:08 2011 +0100
Wibble in tickyDynAlloc (only affects -ticky)
Fall-out from codegen refactoring, undiscovered because
we don't usually build with -ticky
>---
compiler/codeGen
Hello,
I've been working on the new code generator in GHC, and ran across a
curious hunk in a commit from you back from 2008. In "Merging in the new
codegen branch", you update IdLabels to contain information about whether
or not they have CAFs. You then fill in this information in the
following
...@haskell.org] On
| Behalf Of Simon Marlow
| Sent: 04 November 2009 17:03
| To: cvs-ghc@haskell.org
| Subject: patch applied (ghc): Finish #3439: -ticky implies -debug at link
| time; the ticky "way" has gone
|
| Wed Nov 4 06:55:07 PST 2009 Simon Marlow
| * Finish #3439: -ticky implies -debug at
Wed Nov 4 06:55:07 PST 2009 Simon Marlow
* Finish #3439: -ticky implies -debug at link time; the ticky "way" has gone
Ignore-this: 9ca68e63bbc34c8f6f216efe27e259eb
To get ticky profiling you still have to compile with -ticky (for
those modules that you want to profile), b
Tue Oct 20 08:52:01 PDT 2009 simo...@microsoft.com
* Allow -ticky and -prof together
Ignore-this: cddba709acd29968e0517a398e99b49c
The two used to be incompatible, but they aren't any longer.
In fact, -ticky should not be a 'way' any more, and doing that
is on Simo
Thu Oct 8 09:27:52 PDT 2009 simo...@microsoft.com
* Make ghci work with libraries compiled with -ticky
Ignore-this: 97d2d10fbc5fa21afb1076920c1aec66
This is a follow up to the patch tha fixes Trac #3439.
We had forgotten the dynamic linker, which needs to
know all these ticky
Thu Oct 8 09:27:52 PDT 2009 simo...@microsoft.com
* Make ghci work with libraries compiled with -ticky
Ignore-this: 97d2d10fbc5fa21afb1076920c1aec66
This is a follow up to the patch tha fixes Trac #3439.
We had forgotten the dynamic linker, which needs to
know all these ticky
Fri Sep 18 04:46:31 PDT 2009 Simon Marlow
* Fix #3439: -debug implies -ticky, and -ticky code links with any RTS
Ignore-this: 2bc27f861f33a5db471539b0d13618b4
M ./compiler/main/Packages.lhs -3 +11
M ./includes/rts/Config.h -1 +2
M ./includes/stg/Ticky.h -3
M ./rts/Ticky.c
Sent: 19 August 2009 18:48
| To: Simon Peyton-Jones
| Subject: Re: ticky
|
| On Wed, Aug 19, 2009 at 04:03:16PM +0100, Ian Lynagh wrote:
|> On Wed, Aug 19, 2009 at 03:53:37PM +0100, Ian Lynagh wrote:
|> >
|> > Validating with this in mk/validate.mk:
|> >
|> > GhcLib
Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 21/08/2009 10:04, Simon Peyton-Jones wrote:
| > I've cross-linked. Probably
| http://hackage.haskell.org/trac/ghc/wiki/DebuggingGhcCrashes should be under
| "Building", but I have not moved it.
|
| Why under Building?
August 2009 16:33
| To: Simon Peyton-Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 20/08/2009 14:43, Simon Peyton-Jones wrote:
|> Thanks Ian
|>
|> I have started
|> http://hackage.haskell.org/trac/ghc/wiki/Building/DebuggingGHC
|> (referenced from the Building Gui
s
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 20/08/2009 14:43, Simon Peyton-Jones wrote:
| > Thanks Ian
| >
| > I have started
| > http://hackage.haskell.org/trac/ghc/wiki/Building/DebuggingGHC
| > (referenced from the Building Guide) to describe debugging strategies for
18:48
| To: Simon Peyton-Jones
| Subject: Re: ticky
|
| On Wed, Aug 19, 2009 at 04:03:16PM +0100, Ian Lynagh wrote:
|> On Wed, Aug 19, 2009 at 03:53:37PM +0100, Ian Lynagh wrote:
|> >
|> > Validating with this in mk/validate.mk:
|> >
|> > GhcLibOpts
: Ian Lynagh [mailto:ig...@earth.li]
| Sent: 19 August 2009 18:48
| To: Simon Peyton-Jones
| Subject: Re: ticky
|
| On Wed, Aug 19, 2009 at 04:03:16PM +0100, Ian Lynagh wrote:
| > On Wed, Aug 19, 2009 at 03:53:37PM +0100, Ian Lynagh wrote:
| > >
| > > Validating with this in
Thu Aug 6 02:30:21 PDT 2009 Simon Marlow
* Fix ticky build
Ignore-this: dbabf96061763e3a6cb7259e3ea71bed
M ./rts/Ticky.c -1
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20090806093021-12142-3502a6faaab6616d228b6108a58051b9fecfcc52.gz
Thu Dec 18 08:19:28 PST 2008 Ian Lynagh
* Use DynFlags to work out if we are doing ticky ticky profiling
We used to use StaticFlags
M ./compiler/codeGen/CgClosure.lhs -2 +4
M ./compiler/codeGen/CgTailCall.lhs -1 +2
M ./compiler/codeGen/CgTicky.hs -3 +5
M ./compiler/codeGen
Andrew
Did you make any progress with this? Or are you stuck?
It used to be the case that you had to compile *all* code with -ticky-ticky.
But now it's a per-module flag; we just don't accumulate counts for non-ticky
modules. I often use this when doing perf-debugging of a sin
Fri Apr 25 07:04:34 PDT 2008 Ian Lynagh <[EMAIL PROTECTED]>
* Fix the ticky ticky build
Include TickyCounters.h in Stg.h if we are doing Ticky Ticky.
M ./includes/Stg.h +4
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20080425140434
Mon Apr 30 22:37:19 PDT 2007 Tim Chevalier <[EMAIL PROTECTED]>
* Remove references to -fticky-ticky flag
-ticky is the command-line flag for ticky-ticky profiling now, but
internally, the -fticky-ticky flag was used to represent whether ticky
was on. This led to link errors if th
I've pushed a patch that fixes the segfault that (like I discussed
before) was being caused by passing in -rstderr for the ticky file
option.
So with that patch, and the patch that Simon just pushed, I think
ticky should be OK. As far as I can tell.
One thing I noticed: the compiler optio
Thu Apr 26 12:18:57 PDT 2007 Tim Chevalier <[EMAIL PROTECTED]>
* Avoid segfault when ticky file argument is stderr
If you compiled a program with -ticky and ran it with:
./foo +RTS -rstderr -RTS
the result would be a segfault. This was because the RTS interprets stderr to
mea
Simon Peyton Jones wrote:
Really there should be some complaint from the code generators, not simply
a seg fault.
There is now such a check:
Wed Apr 4 16:50:01 BST 2007 Simon Marlow <[EMAIL PROTECTED]>
* add a safety net to avoid generating bogus code
Cheers,
Simon
__
Oops sorry. I recorded but did not push. Try now.
S
| -Original Message-
| From: Tim Chevalier [mailto:[EMAIL PROTECTED]
| Sent: 25 April 2007 05:43
| To: Simon Peyton-Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 4/24/07, Simon Peyton-Jones <[EMAIL PROTECTED]>
Tue Apr 24 06:30:11 PDT 2007 [EMAIL PROTECTED]
* Make ticky work, at least partly, on 64-bit machines
The ticky StgEntCounter structure was trying to be clever by using a
fixed-width 32-bit field for the registeredp value. But the code generators
are not up to handling structures
On 4/24/07, Simon Peyton-Jones <[EMAIL PROTECTED]> wrote:
Tim
OK, I committed my small patch to ticky. Over to you!
Are you sure you checked in the patch? I don't see any changes to
CgTicky since April 4.
Cheers,
Tim
--
Tim Chevalier * [EMAIL PROTECTED] * Often in error, nev
Tim
OK, I committed my small patch to ticky. Over to you!
Simon
| -Original Message-
| From: Tim Chevalier [mailto:[EMAIL PROTECTED]
| Sent: 06 April 2007 05:04
| To: Simon Peyton-Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 4/5/07, Tim Chevalier <[EMAIL PROTEC
From: Tim Chevalier [mailto:[EMAIL PROTECTED]
| Sent: 06 April 2007 05:04
| To: Simon Peyton-Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 4/5/07, Tim Chevalier <[EMAIL PROTECTED]> wrote:
| > Also -- I just noticed that my segfault isn't the same as your
| > segfault,
On 4/5/07, Tim Chevalier <[EMAIL PROTECTED]> wrote:
Also -- I just noticed that my segfault isn't the same as your
segfault, apparently, because I'm not using a 64-bit machine (or
rather, I am, but my OS thinks it's running on a 32-bit machine for
some reason I don't want to explore). I get a seg
On 4/5/07, Tim Chevalier <[EMAIL PROTECTED]> wrote:
Okay, I'm just trying to reproduce what you did here... in the typedef
for StgEntCounter in Rts.h, I changed the three instances of StgWord32
to StgWord, and correspondingly, in emitTickyCounter in CgTicky.hs, I
changed the three instances of I
On 4/5/07, Simon Peyton-Jones <[EMAIL PROTECTED]> wrote:
more info...
The seg fault was because I was on a 64-bit machine. StgEntCounter (defined in
Rts.h) used some 32-bit fields. The asm code emitted by CgTicky to
statically-allocate the record put the 32-bit words end to end, followed the
Simon
| -Original Message-
| From: Tim Chevalier [mailto:[EMAIL PROTECTED]
| Sent: 04 April 2007 19:21
| To: Simon Peyton-Jones
| Cc: cvs-ghc@haskell.org
| Subject: Re: ticky
|
| On 4/4/07, Simon Peyton-Jones <[EMAIL PROTECTED]> wrote:
| > Do you think ticky is working in the
On 4/4/07, Simon Peyton-Jones <[EMAIL PROTECTED]> wrote:
Do you think ticky is working in the HEAD?
Well, apparently not! (I swear, it was working when I checked it in...)
Any chance you could patch it up?
I'll look at it sometime over the next couple days and if I can't
Hi Kirsten
I'm getting this when compiling the RTS with -ticky:
../compiler/ghc-inplace -H32m -W -fno-warn-unused-matches -fwarn-unused-imports
-optc-O2 -static -I. -#include HCIncludes.h -fvia-C -dcmm-lint -hisuf t_hi
-hcsuf t_hc -osuf t_o -ticky -#include posix/Itimer.h -c PrimOps.c
Wed Apr 4 07:10:13 PDT 2007 Simon Marlow <[EMAIL PROTECTED]>
* add comments about ticky RTS ways
M ./mk/config.mk.in +2
___
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
Fri Feb 9 06:08:18 PST 2007 Simon Marlow <[EMAIL PROTECTED]>
* Ticky is an RTS-only way; also fix collateral damage to other ways
The ticky static flag was being poked too early, which lead to
breakage in the -prof way amongst other things. I've installed some
sanity check
Wed Feb 7 00:45:27 PST 2007 Kirsten Chevalier <[EMAIL PROTECTED]>
* Update to manual section for ticky-ticky
I added a slightly more helpful remark in the profiling
section of the manual about how to use ticky-ticky.
M ./docs/users_guide/profiling.xml
Wed Feb 7 00:14:04 PST 2007 Kirsten Chevalier <[EMAIL PROTECTED]>
* Lightweight ticky-ticky profiling
The following changes restore ticky-ticky profiling to functionality
from its formerly bit-rotted state. Sort of. (It got bit-rotted as part
of the switch to the C-- ba
37 matches
Mail list logo