[PATCH] Explicitly specify CMake MSVC stack size

Greg_Bedwell at sn.scee.net Greg_Bedwell at sn.scee.net
Tue Sep 24 23:31:00 PDT 2013


Thanks.  Please could you commit for me?  I don't have access yet.


Rafael EspĂ­ndola <rafael.espindola at gmail.com> wrote on 24/09/2013 

> From: Rafael EspĂ­ndola <rafael.espindola at gmail.com>
> To: Greg_Bedwell at sn.scee.net
> Cc: llvm cfe <cfe-commits at cs.uiuc.edu>, llvm-commits <llvm-
> commits at cs.uiuc.edu>
> Date: 24/09/2013 22:15
> Subject: Re: [PATCH] Explicitly specify CMake MSVC stack size
> On 24 September 2013 17:13,  <Greg_Bedwell at sn.scee.net> wrote:
> > Thanks for looking at this!
> >
> >> This changes the stack of every tool, not just clang, right?
> >
> > Yes and no depending on how you look at it.  It increases the stack 
> > for every tool for anybody that is already using CMake 2.8.11, but it
> > prevents the stack size from changing for anybody in future that is 
> > using a previous version.  According to cmake.org, 2.8.11 was released 
> > May 2013, so I'm guessing that a lot of people may not have had reason 
> > update yet.  The only reason I updated was because the previous 
version I
> > was using was too old for compiler-rt which I wanted to mess around 
> >
> >>Is it possible to change it just for clang?
> >>
> >
> > It's true that the only place I've actually observed us actually 
> > 10MB of stack is in clang.exe, but then I have more exhaustive tests 
> > clang.exe than any other tool :).  The primary motivation here is to 
> > the situation where two people building the exact same revision with
> > exactly the same MSVC version get different results with one of the 
> > depending on the version of CMake they happen to have installed.  It 
> > feels like it's a horrible day of debugging for someone later on down 
> > line.
> >
> > My idea was that we apply this to every tool for now in order to 
> > previous behaviour, and then at the point where we mandate a version 
> > CMake that supports CMAKE_CXX_STACK_SIZE we can find more appropriate
> > values for each tool (most likely, the default 1MB I'll concede for 
> > other than clang).  This will then guarantee that anyone trying to 
> > down any strange regressions can at least track it down to a single
> > revision that changed the stack size.
> >
> > Does this sound like a reasonable approach?
> I think so.
> Cheers,
> Rafael

This email and any files transmitted with it are confidential and intended 
solely for the use of the individual or entity to whom they are addressed. 
If you have received this email in error please notify postmaster at scee.net
This footnote also confirms that this email message has been checked for 
all known viruses.
Sony Computer Entertainment Europe Limited
Registered Office: 10 Great Marlborough Street, London W1F 7LP, United 
Registered in England: 3277793

P Please consider the environment before printing this e-mail

More information about the llvm-commits mailing list