Re: [Chicken-hackers] [PATCH] proposed bugfix for #706

[Alan Post]

I can't see buffer2 being deallocated.  Is there a free() missing
here?

-Alan

On Wed, Sep 28, 2011 at 05:46:04PM +0200, Christian Kellermann wrote:
> Hi,
> 
> attached is a patch as a proposed fix for bug Ã#706.
> 
> This will need some discussion as I am not sure what I am doing here :)
> 
> First of all it introduces:
> 
> 1. A new primitive C_file_exists that will use access to check wether we 
> could access a file
> 2. Reroutes the file-exists? procedure to use C_file_exists
> 3. A new error code for barf() to signal system call errors
> 
> Ad 1.:
> 
> Thomas Chust suggested this and the code seems a bit easier and we
> do not need to allocate all that memory we aren't interested in.
> AÃlso we dodge the overflow situation altogether for this small use
> case.
> 
> We will signal an error if we get anything negative besides ENOENT back.
> 
> I am not sure if I did the name mangling right. Also where does
> "buffer" come from? Please investigate!
> 
> Ad 2:
> 
>  trivial
> 
> Ad 3:
> 
> To signal the error above I wanted to construct a proper condition
> (exn i/o) or something but I am unsure how to do this. To base the
> discussion upon facts, I have created a new error type for barf so
> at lease I get a little more helpful hint at what's going on.
> 
> Looking forward hearing your opinions,
> 
> Christian
> 
> 
> -- 
> Who can (make) the muddy water (clear)? Let it be still, and it will
> gradually become clear. Who can secure the condition of rest? Let
> movement go on, and the condition of rest will gradually arise.
>  -- Lao Tse. 

> From 341ac1c69df18eac9a5ebf0cfc5553fa62bff3bd Mon Sep 17 00:00:00 2001
> From: Christian Kellermann <address@hidden>
> Date: Wed, 28 Sep 2011 17:23:57 +0200
> Subject: [PATCH] Replace ##sys#file-info with ##sys#file-exists
> 
> This also introduces a simpler runtime function that calls access to
> test for file accessability.
> 
> This fixes bug #706 reported by Sven Hartrumpf.
> 
> The old version  returned false whenever fstat failed.  In Sven's case
> the size parameter overflowed, and file-exists? returns #f.
> 
> Thanks to Thomas Chust for the suggestion.
> ---
>  chicken.h   |    3 ++-
>  library.scm |    3 ++-
>  runtime.c   |   31 +++++++++++++++++++++++++++++++
>  3 files changed, 35 insertions(+), 2 deletions(-)
> 
> diff --git a/chicken.h b/chicken.h
> index 8c6eff3..47b12f7 100644
> --- a/chicken.h
> +++ b/chicken.h
> @@ -572,7 +572,7 @@ static inline int isinf_ld (long double x)
>  #define C_CIRCULAR_DATA_ERROR                         36
>  #define C_BAD_ARGUMENT_TYPE_NO_BOOLEAN_ERROR          37
>  #define C_BAD_ARGUMENT_TYPE_NO_LOCATIVE_ERROR         38
> -
> +#define C_SYSCALL_ERROR                               39
>  
>  /* Platform information */
>  #if defined(C_BIG_ENDIAN)
> @@ -1674,6 +1674,7 @@ C_fctexport void C_ccall C_make_tagged_pointer(C_word 
> c, C_word closure, C_word
>  C_fctexport void C_ccall C_ensure_heap_reserve(C_word c, C_word closure, 
> C_word k, C_word n) C_noret;
>  C_fctexport void C_ccall C_return_to_host(C_word c, C_word closure, C_word 
> k) C_noret;
>  C_fctexport void C_ccall C_file_info(C_word c, C_word closure, C_word k, 
> C_word port) C_noret;
> +C_fctexport void C_ccall C_file_exists(C_word c, C_word closure, C_word k, 
> C_word port) C_noret;
>  C_fctexport void C_ccall C_get_environment_variable(C_word c, C_word 
> closure, C_word k, C_word name) C_noret;
>  C_fctexport void C_ccall C_get_symbol_table_info(C_word c, C_word closure, 
> C_word k) C_noret;
>  C_fctexport void C_ccall C_get_memory_info(C_word c, C_word closure, C_word 
> k) C_noret;
> diff --git a/library.scm b/library.scm
> index e07e9c4..5fef68e 100644
> --- a/library.scm
> +++ b/library.scm
> @@ -191,6 +191,7 @@ EOF
>  (define ##sys#call-host (##core#primitive "C_return_to_host"))
>  (define return-to-host ##sys#call-host)
>  (define ##sys#file-info (##core#primitive "C_file_info"))
> +(define ##sys#file-exists (##core#primitive "C_file_exists"))
>  (define ##sys#symbol-table-info (##core#primitive "C_get_symbol_table_info"))
>  (define ##sys#memory-info (##core#primitive "C_get_memory_info"))
>  (define (current-milliseconds) (##core#inline_allocate 
> ("C_a_i_current_milliseconds" 4) #f))
> @@ -1979,7 +1980,7 @@ EOF
>    (##sys#pathname-resolution
>      name
>      (lambda (name)
> -      (and (##sys#file-info (##sys#platform-fixup-pathname name)) name) )
> +      (and (##sys#file-exists (##sys#platform-fixup-pathname name)) name) )
>      #:exists?) )
>  
>  (define (directory-exists? name)
> diff --git a/runtime.c b/runtime.c
> index c0c91bc..32804cf 100644
> --- a/runtime.c
> +++ b/runtime.c
> @@ -737,6 +737,7 @@ static C_PTABLE_ENTRY *create_initial_ptable()
>    C_pte(C_ensure_heap_reserve);
>    C_pte(C_return_to_host);
>    C_pte(C_file_info);
> +  C_pte(C_file_exists);
>    C_pte(C_get_symbol_table_info);
>    C_pte(C_get_memory_info);
>    C_pte(C_decode_seconds);
> @@ -1616,6 +1617,11 @@ void barf(int code, char *loc, ...)
>      c = 0;
>      break;
>  
> +  case C_SYSCALL_ERROR:
> +    msg = C_text("Underlying syscall error");
> +    c = 1;
> +    break;
> +
>    default: panic(C_text("illegal internal error code"));
>    }
>    
> @@ -7776,6 +7782,31 @@ void C_ccall C_file_info(C_word c, C_word closure, 
> C_word k, C_word name)
>    file_info_2(NULL);
>  }
>  
> +void C_ccall C_file_exists(C_word c, C_word closure, C_word k, C_word name)
> +{
> +  C_word v = C_SCHEME_FALSE;
> +  int call_res = 0;
> +
> +  int len = C_header_size(name);
> +  char *buffer2;
> +  int res = 0;
> +
> +  buffer2 = buffer;
> +  if(len >= sizeof(buffer)) {
> +    if((buffer2 = (char *)C_malloc(len + 1)) == NULL)
> +      barf(C_OUT_OF_MEMORY_ERROR, "stat");
> +  }
> +  C_strncpy(buffer2, C_c_string(name), len);
> +  buffer2[ len ] = '\0';
> +  call_res=access(buffer2, F_OK);
> +  if (call_res == 0) {
> +    v = C_SCHEME_TRUE;
> +  } else if (errno != ENOENT) {
> +    barf(C_SYSCALL_ERROR, "access");
> +  }
> +
> +  C_kontinue(k, v);
> +}
>  
>  void file_info_2(void *dummy)
>  {
> -- 
> 1.7.4.1
> 

> _______________________________________________
> Chicken-hackers mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/chicken-hackers


-- 
.i ma'a lo bradi cu penmi gi'e du