jeff@cjsa.WA.COM (Jeffery Small) (11/03/88)
When compiling a program on an AT&T 3B1 (running SystemV.2 +/-), I am
getting the following warning message from lint when I use malloc(3):
warning: possible pointer alignment problem
Here is some sample code to illustrate the problem:
------- cut here -------------------------------------------------------
#define NULL 0
typedef struct {
unsigned int x;
unsigned int y;
unsigned char *z;
} X;
main()
{
X *getptr();
if ( getptr() == (X *) NULL)
return (1);
else
return(0);
/*NOTREACHED*/
}
X *getptr()
{
char *malloc();
return( (X *) malloc(sizeof(X)) ); /* WARNING FOR THIS LINE */
}
------ end -----------------------------------------------------------
Now, since the man page for malloc reads:
Malloc returns a pointer to a block of at least size bytes
suitably aligned for any use.
I assume that lint is complaining not about the pointer returned from
malloc(), but instead, about the alignment of the *z pointer which is
a member of the structure.
The program works on this machine, and while I know that lint is warning
about a portability problem to other architectures, I do not know what I
need to do to satisfy lint. Is some manual padding required within the
structure, and if so, what are the rules for this.
BTW, I have tried a number of different arrangements of the structure
members with no satisfactory results.
Thanks for any pointers. (Yea, I know :-)
--
Jeffery Small (206) 485-5596 uw-beaver!uw-nsr!uw-warp
C. Jeffery Small and Associates !cjsa!jeff
19112 152nd Ave NE - Woodinville, WA 98072 uunet!nwnexusdaveh@marob.MASA.COM (Dave Hammond) (11/03/88)
In article <142@cjsa.WA.COM> jeff@cjsa.WA.COM (Jeffery Small) writes: >When compiling a program on an AT&T 3B1 (running SystemV.2 +/-), I am >getting the following warning message from lint when I use malloc(3): > > warning: possible pointer alignment problem >[...sample code deleted...] The program code is not at fault. The problem is that lint does not know about malloc returning a suitably aligned pointer, therefore it complains. You could do some fancy declaration footwork and declare X *malloc(); for the file, but be careful that there are no instances of any other malloc casts (including char *) within the same file. Most folks just ignore this warning when issued for malloc. Dave Hammond UUCP: ...!uunet!masa.com!{marob,dsix2}!daveh DOMAIN: daveh@marob.masa.com ----------------------------------------------------------------------------
crossgl@ingr.UUCP (Gordon Cross) (11/03/88)
In article <142@cjsa.WA.COM>, jeff@cjsa.WA.COM (Jeffery Small) writes: > When compiling a program on an AT&T 3B1 (running SystemV.2 +/-), I am > getting the following warning message from lint when I use malloc(3): > > warning: possible pointer alignment problem The warning that lint is giving you is based on the fact that you are casting a character pointer (returned from malloc) into another pointer type having more stringent alignment requirements. Malloc does guarantee that all pointers it returns are properly aligned to use with any data type but lint does not assume this since "malloc" is just another function (you could write your own version you know). Your best bet is to just ignore the warning since you are NOT in reality going to have any problems from this! Gordon Cross Intergraph Corp. Huntsville, AL
guy@auspex.UUCP (Guy Harris) (11/04/88)
>I assume that lint is complaining not about the pointer returned from >malloc(), but instead, about the alignment of the *z pointer which is >a member of the structure. Incorrect assumption. "lint" doesn't know "malloc()" from a hole in the ground; it has no idea that it returns maximally-aligned pointers. This might be considered a deficiency in "lint". There is nothing wrong with the alignment of the "*z pointer". Do not adjust your set. As one version of the "lint" manual page said, under the BUGS section, "There are some things you just *can't* get 'lint' to shut up about."
steve@umigw.MIAMI.EDU (steve emmerson) (11/04/88)
I sometimes use:
#ifdef lint
# define malloc(s) 0
#else
extern char *malloc();
#endif
Unfortunately, this means that lint(1) can no longer check the type of the
argument. C'est la vie.
--
Steve Emmerson Inet: steve@umigw.miami.edu [128.116.10.1]
SPAN: miami::emmerson (host 3074::) emmerson%miami.span@star.stanford.edu
UUCP: ...!ncar!umigw!steve emmerson%miami.span@vlsi.jpl.nasa.gov
"Computers are like God in the Old Testament: lots of rules and no mercy"tonyw@microsoft.UUCP (Tony Williams) (11/05/88)
In article <355@marob.MASA.COM> daveh@marob.masa.com (Dave Hammond) writes: |In article <142@cjsa.WA.COM> jeff@cjsa.WA.COM (Jeffery Small) writes: |>When compiling a program on an AT&T 3B1 (running SystemV.2 +/-), I am |>getting the following warning message from lint when I use malloc(3): |> |> warning: possible pointer alignment problem |>[...sample code deleted...] | |The program code is not at fault. The problem is that lint does not |know about malloc returning a suitably aligned pointer, therefore |it complains. You could do some fancy declaration footwork and declare | |X *malloc(); DO NOT DO THIS. Since you are using lint, I assume you care somewhat about portability :-). There are machines out there for which struct X * and char * have different representations. On these machines, malloc will return the char* representation and you must have the cast to convert it. For the same reason, you must have a declaration extern char* malloc(); in your programs, so that the compiler does not assume it is int. I am afraid that we just have to live with some features of lint. Tony Williams
bill@twwells.uucp (T. William Wells) (11/07/88)
In article <182@umigw.MIAMI.EDU> steve@umigw.miami.edu (steve emmerson) writes:
: I sometimes use:
:
: #ifdef lint
: # define malloc(s) 0
: #else
: extern char *malloc();
: #endif
:
: Unfortunately, this means that lint(1) can no longer check the type of the
: argument. C'est la vie.
But you could do this:
#ifdef lint
static void malloc_kludge(n) size_t n; { n = n; }
#define malloc(x) (malloc_kludge(x), 0)
#else
extern char *malloc();
#endif
---
Bill
{uunet|novavax}!proxftl!twwells!billmeissner@xyzzy.UUCP (Usenet Administration) (11/07/88)
In article <355@marob.MASA.COM> daveh@marob.masa.com (Dave Hammond) writes: /* stuff about lint giving errors calling malloc deleted */ | The program code is not at fault. The problem is that lint does not | know about malloc returning a suitably aligned pointer, therefore | it complains. You could do some fancy declaration footwork and declare | | X *malloc(); No, No, No, No, No, No! There are implementations of C on word-based machines, which use different representations for different types of pointers. Since the library routine malloc returns a char *, and you have declared to return X *, you will get undefined behvior. This is NOT guaranteed by K&R, H&S, or X3J11. -- Michael Meissner, Data General. Uucp: ...!mcnc!rti!xyzzy!meissner Arpa: meissner@dg-rtp.DG.COM (or) meissner%dg-rtp.DG.COM@relay.cs.net
daveh@marob.MASA.COM (Dave Hammond) (11/07/88)
In article <1169@microsoft.UUCP> tonyw@microsoft.UUCP (Tony Williams) writes: >In article <355@marob.MASA.COM> daveh@marob.masa.com (Dave Hammond) writes: >| You could do some fancy declaration footwork and declare >| >|X *malloc(); > >DO NOT DO THIS. Since you are using lint, I assume you care somewhat >about portability :-). There are machines out there for which >struct X * and char * have different representations. >On these machines, malloc will return the char* representation and you >must have the cast to convert it. Sorry! My example should have shown conditional inclusion for lint only, e.g.: #ifdef lint X *malloc(); #else char *malloc(); #endif Dave Hammond UUCP: ...!uunet!masa.com!{marob,dsix2}!daveh DOMAIN: daveh@marob.masa.com ----------------------------------------------------------------------------