A parsing and analysis library for the C language ()

Pekka Enberg | 1 Feb 10:55

[RFC/PATCH 2/2] sparse, llvm: Fix varargs functions

From: Benjamin Herrenschmidt <benh <at> kernel.crashing.org>

We need to tell llvm about it or it won't generate the proper
stack frame & argument list on some architectures.

Cc: Christopher Li <sparse <at> chrisli.org>
Cc: Jeff Garzik <jgarzik <at> redhat.com>
Cc: Linus Torvalds <torvalds <at> linux-foundation.org>
Signed-off-by: Benjamin Herrenschmidt <benh <at> kernel.crashing.org>
[ penberg <at> kernel.org: Fix function pointer calls ]
Signed-off-by: Pekka Enberg <penberg <at> kernel.org>
---
 sparse-llvm.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/sparse-llvm.c b/sparse-llvm.c
index a291a0d..9226a21 100644
--- a/sparse-llvm.c
+++ b/sparse-llvm.c
@@ -79,7 +79,7 @@ static LLVMTypeRef sym_func_type(LLVMModuleRef module, struct symbol *sym)
 		arg_type[idx++] = symbol_type(module, arg_sym);
 	} END_FOR_EACH_PTR(arg);
 	func_type = LLVMFunctionType(ret_type, arg_type, n_arg,
-				     /* varargs? */ 0);
+				     sym->ctype.base_type->variadic);

 	return func_type;
 }
@@ -744,7 +744,7 @@ static LLVMTypeRef get_func_type(struct function *fn, struct instruction *insn)
 	} END_FOR_EACH_PTR(arg);

(Continue reading)

Pekka Enberg | 1 Feb 10:55

[RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

From: Linus Torvalds <torvalds <at> linux-foundation.org>

On Tue, Aug 30, 2011 at 10:43 AM, Jeff Garzik <jeff <at> garzik.org> wrote:
> * if someone knows how to access a function declaration, I can solve the
> varargs problem

Hmm. Right now we do not have access to the function declaration at
linearize time. We've checked that the arguments match, and we've cast
the arguments to the right types (evaluate.c), so the thinking was
that you just use the arguments as-is.

But if llvm needs the declaration of a function, we'd need to squirrel it away.

Cc: Benjamin Herrenschmidt <benh <at> kernel.crashing.org>
Cc: Christopher Li <sparse <at> chrisli.org>
Cc: Jeff Garzik <jgarzik <at> redhat.com>
Cc: Linus Torvalds <torvalds <at> linux-foundation.org>
[ penberg <at> kernel.org: Fix validation/context.c breakage. ]
Signed-off-by: Pekka Enberg <penberg <at> kernel.org>
---
 linearize.c |    8 ++++++++
 linearize.h |    1 +
 2 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/linearize.c b/linearize.c
index 1899978..7d57474 100644
--- a/linearize.c
+++ b/linearize.c
@@ -1195,6 +1195,7 @@ static pseudo_t linearize_call_expression(struct entrypoint *ep, struct expressi
 	struct instruction *insn = alloc_typed_instruction(OP_CALL, expr->ctype);

(Continue reading)

Benjamin Herrenschmidt | 1 Feb 11:47

Re: [RFC/PATCH 2/2] sparse, llvm: Fix varargs functions

On Wed, 2012-02-01 at 11:55 +0200, Pekka Enberg wrote:
> From: Benjamin Herrenschmidt <benh <at> kernel.crashing.org>
> 
> We need to tell llvm about it or it won't generate the proper
> stack frame & argument list on some architectures.
> 
> Cc: Christopher Li <sparse <at> chrisli.org>
> Cc: Jeff Garzik <jgarzik <at> redhat.com>
> Cc: Linus Torvalds <torvalds <at> linux-foundation.org>
> Signed-off-by: Benjamin Herrenschmidt <benh <at> kernel.crashing.org>
> [ penberg <at> kernel.org: Fix function pointer calls ]
> Signed-off-by: Pekka Enberg <penberg <at> kernel.org>
> ---
>  sparse-llvm.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/sparse-llvm.c b/sparse-llvm.c
> index a291a0d..9226a21 100644
> --- a/sparse-llvm.c
> +++ b/sparse-llvm.c
> @@ -79,7 +79,7 @@ static LLVMTypeRef sym_func_type(LLVMModuleRef module, struct symbol *sym)
>  		arg_type[idx++] = symbol_type(module, arg_sym);
>  	} END_FOR_EACH_PTR(arg);
>  	func_type = LLVMFunctionType(ret_type, arg_type, n_arg,
> -				     /* varargs? */ 0);
> +				     sym->ctype.base_type->variadic);
>  
>  	return func_type;
>  }

(Continue reading)

Jeff Garzik | 1 Feb 14:58

Re: [RFC/PATCH 2/2] sparse, llvm: Fix varargs functions

On 02/01/2012 05:47 AM, Benjamin Herrenschmidt wrote:
> On Wed, 2012-02-01 at 11:55 +0200, Pekka Enberg wrote:
>> From: Benjamin Herrenschmidt<benh <at> kernel.crashing.org>
>>
>> We need to tell llvm about it or it won't generate the proper
>> stack frame&  argument list on some architectures.
>>
>> Cc: Christopher Li<sparse <at> chrisli.org>
>> Cc: Jeff Garzik<jgarzik <at> redhat.com>
>> Cc: Linus Torvalds<torvalds <at> linux-foundation.org>
>> Signed-off-by: Benjamin Herrenschmidt<benh <at> kernel.crashing.org>
>> [ penberg <at> kernel.org: Fix function pointer calls ]
>> Signed-off-by: Pekka Enberg<penberg <at> kernel.org>
>> ---
>>   sparse-llvm.c |    4 ++--
>>   1 files changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/sparse-llvm.c b/sparse-llvm.c
>> index a291a0d..9226a21 100644
>> --- a/sparse-llvm.c
>> +++ b/sparse-llvm.c
>> @@ -79,7 +79,7 @@ static LLVMTypeRef sym_func_type(LLVMModuleRef module, struct symbol *sym)
>>   		arg_type[idx++] = symbol_type(module, arg_sym);
>>   	} END_FOR_EACH_PTR(arg);
>>   	func_type = LLVMFunctionType(ret_type, arg_type, n_arg,
>> -				     /* varargs? */ 0);
>> +				     sym->ctype.base_type->variadic);
>>
>>   	return func_type;
>>   }

(Continue reading)

Christopher Li | 2 Feb 01:09

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On Wed, Feb 1, 2012 at 1:55 AM, Pekka Enberg <penberg <at> kernel.org> wrote:
> From: Linus Torvalds <torvalds <at> linux-foundation.org>
>
> On Tue, Aug 30, 2011 at 10:43 AM, Jeff Garzik <jeff <at> garzik.org> wrote:
>> * if someone knows how to access a function declaration, I can solve the
>> varargs problem
>
> Hmm. Right now we do not have access to the function declaration at
> linearize time. We've checked that the arguments match, and we've cast
> the arguments to the right types (evaluate.c), so the thinking was
> that you just use the arguments as-is.

Ok, this patch definitely works. However, I think there is ways to get the
type without this patch. The more general question is, how to get type of
any given pseudo register. Currently it is a bit twisted but that information
should be accessible. I believe that has been discuss on the mail list
before. That is why we have insn->type. LLVM want to know every type
of every value in the back end any way. If we are going to do proper
llvm style "get element pointer" in the back end, we need to access type
of every pesudo register.

The current twisted way to get type from the pseudo register is some thing like
this (totally untested code):

swtich (pseudo->type) {
     case PSEDUO_REG:
          return pseudo->def->type;
     case PSEUDO_ARG:
     case PSEUDO_SYMBOL:
          return psuedo->symbol;

(Continue reading)

Benjamin Herrenschmidt | 2 Feb 01:54

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On Wed, 2012-02-01 at 16:09 -0800, Christopher Li wrote:
> 
> Ok, this patch definitely works. However, I think there is ways to get the
> type without this patch. The more general question is, how to get type of
> any given pseudo register. Currently it is a bit twisted but that information
> should be accessible. I believe that has been discuss on the mail list
> before. That is why we have insn->type. LLVM want to know every type
> of every value in the back end any way. If we are going to do proper
> llvm style "get element pointer" in the back end, we need to access type
> of every pesudo register.

We also should probably store the LLVM Ref of the type once "converted"
into the original struct symbol so that we don't have to re-create it or
all the time (thinking especially of typedef'ed function pointers) or
search for it by name (functions).

Cheers,
Ben.

--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Jeff Garzik | 2 Feb 02:03

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On 02/01/2012 07:09 PM, Christopher Li wrote:
> Ok, this patch definitely works. However, I think there is ways to get the
> type without this patch. The more general question is, how to get type of
> any given pseudo register.

That is useful, yes.  But it does not address this specific problem.

We need the function declaration remembered, rather than what we have 
now -- a list of arguments with full type information, specific to its 
callsite.

You cannot deduce that a function call is/not varargs presently, even 
with a working pseudo->type setup.

varargs is just one of those annoying areas where the compiler needs to 
have rather specific knowledge, in order to properly construct a call

	Jeff

--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Benjamin Herrenschmidt | 2 Feb 02:22

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On Wed, 2012-02-01 at 20:03 -0500, Jeff Garzik wrote:
> On 02/01/2012 07:09 PM, Christopher Li wrote:
> > Ok, this patch definitely works. However, I think there is ways to get the
> > type without this patch. The more general question is, how to get type of
> > any given pseudo register.
> 
> 
> That is useful, yes.  But it does not address this specific problem.
> 
> We need the function declaration remembered, rather than what we have 
> now -- a list of arguments with full type information, specific to its 
> callsite.
> 
> You cannot deduce that a function call is/not varargs presently, even 
> with a working pseudo->type setup.
> 
> varargs is just one of those annoying areas where the compiler needs to 
> have rather specific knowledge, in order to properly construct a call

And there could be more. For example specific attributes on the
declaration may affect the ABI for the call (think asmlinkage or good
old pascal calling conventions on old macos 

Not something we absolutely need to sort out right now but another
reason why we really need to base the LLVM side definition based on the
declaration.

I'll try to toy a bit more this week-end see if the patches work for all
cases I can think of. We really have two different things (represented
by the two different hunks) which we might try to better factor:

(Continue reading)

Christopher Li | 2 Feb 02:22

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On Wed, Feb 1, 2012 at 5:03 PM, Jeff Garzik <jeff <at> garzik.org> wrote:
> That is useful, yes.  But it does not address this specific problem.
>
> We need the function declaration remembered, rather than what we have now --
> a list of arguments with full type information, specific to its callsite.
>
> You cannot deduce that a function call is/not varargs presently, even with a
> working pseudo->type setup.

I am not trying do deduce function call/varargs from the call. I am
trying to get
it from call_insn->func->type directly. Similar to Linus' patch,
instead of store fntype
at call_insn->fntype, store it at call_insn->func->type directly. It should work
the similar way as Linus's patch, no?

Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo <at> vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Christopher Li | 2 Feb 02:33

Re: [RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to backend

On Wed, Feb 1, 2012 at 5:22 PM, Benjamin Herrenschmidt
<benh <at> kernel.crashing.org> wrote:
> And there could be more. For example specific attributes on the
> declaration may affect the ABI for the call (think asmlinkage or good
> old pascal calling conventions on old macos 
>
> Not something we absolutely need to sort out right now but another
> reason why we really need to base the LLVM side definition based on the
> declaration.

Yes, that is what I am talking about too. Notice that insn->func is a pseudo
not a symbol. Pseudo can be result from another expression. That is my
suggestion in the counter RFC, put the full symbol type into pseudo->type.
You should have access of that information.

>
> I'll try to toy a bit more this week-end see if the patches work for all
> cases I can think of. We really have two different things (represented
> by the two different hunks) which we might try to better factor:
>
> The case of a function call where we are after the declaration, and the
> case of creating an llvm type containing a function pointer (which can
> happen as part of the first one if an argument is a function pointer)
> for which we are looking at the base_type.
>
> We should at least make it a single piece of code that takes a symbol
> and shoots out a llvm ref.

Do you mean take a pseudo then shoot out a llvm ref? The pseudo is
the more general form, a pseudo can come from symbol node or

(Continue reading)

Group

gmane.comp.parsers.sparse.

Search Archive

Page

1 | 2 | 3

Articles in period: 21

February 2012

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane

Recent entries

[PATCH] Updated __nocast vs __bitwise documentation (10 May 08:54)
simplify: conservative handling of casts with pointers (8 May 16:13)
Fwd: dependency tee from c parser entities downto token (6 May 01:38)
Fwd: dependency tee from c parser entities downto token (5 May 21:56)
[PATCH] unssa: track use of newly added pseudo (3 May 16:59)
dependency tee from c parser entities downto token (24 Apr 11:54)
Please be informed! (5 Apr 08:14)
Killing off __cond_lock() (24 Mar 17:23)
help needed: document __nocast vs __bitwise (Was: [PATCH 00/16] mm: pr (23 Mar 19:17)
Using c2xml on kernel sources (20 Mar 10:00)
Patch: add __builtin_stpcpy and __sync_synchronize to builtin function (15 Mar 15:15)
MAIL 10.03.2012 (11 Mar 11:05)
strings (3 Mar 11:46)
sparse-llvm: pseudo_to_value: Assertion `sym->ident == ((void *)0)' fa (13 Feb 08:17)
[RFC/PATCH 1/2] sparse, llvm: Make function declaration accessible to (1 Feb 10:55)
[PATCH] evaluate: reject post-ops on void* (23 Jan 00:31)
sparse-llvm has been merged (19 Jan 03:00)
[PATCH] sparse: Add 'leaf' to ignored attributes. (17 Jan 23:47)
[PATCH 1/3] sparse, llvm: Use LLVMInt1Type() in sym_basetype_type() (21 Dec 10:25)
Sparse 0.4.4 is released (25 Nov 22:13)
[GIT PULL] Sparse LLVM backend (25 Nov 08:46)

Archives

61	May 2012
7	April 2012
26	March 2012
21	February 2012
10	January 2012
10	December 2011
55	November 2011
19	October 2011
8	September 2011
149	August 2011
1	July 2011
5	June 2011
22	May 2011
49	April 2011
5	March 2011
6	February 2011
1	January 2011
9	December 2010
12	November 2010
71	October 2010
4	September 2010
13	August 2010
15	July 2010
22	June 2010
7	May 2010
24	April 2010
45	March 2010
37	February 2010
7	January 2010
18	December 2009
6	November 2009
20	October 2009
53	September 2009
73	August 2009
75	July 2009
27	June 2009
7	May 2009
44	April 2009
68	March 2009
44	February 2009
21	January 2009
116	December 2008
23	November 2008
41	September 2008
9	August 2008
46	July 2008
52	June 2008
44	May 2008
121	April 2008
36	March 2008
20	February 2008
8	January 2008
18	December 2007
13	November 2007
34	October 2007
15	September 2007
22	August 2007
132	July 2007
160	June 2007
159	May 2007
16	April 2007
96	March 2007
108	February 2007
48	January 2007
67	December 2006
39	November 2006
40	October 2006
25	September 2006
17	August 2006
16	July 2006
6	June 2006
9	May 2006
7	April 2006
12	March 2006
13	January 2006
16	December 2005

Design

Zawodny | Right Menu | Left Menu

Your Own Design

Paste the URL of your CSS below. Download CSS template