| Message ID | 20230614142855.1535558-1-siddhesh@sourceware.org |
|---|---|
| State | New |
| Headers | show |
| Series | manual: Document deviation of realloc from C2x/N2464 | expand |
* Siddhesh Poyarekar: > Explicitly note that the realloc behaviour chosen in glibc is deliberate > and well defined and suggest that strictly compliant code should avoid > passing zero to realloc. > > Signed-off-by: Siddhesh Poyarekar <siddhesh@sourceware.org> > --- > manual/memory.texi | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/manual/memory.texi b/manual/memory.texi > index 5781a64f35..7697a10f72 100644 > --- a/manual/memory.texi > +++ b/manual/memory.texi > @@ -832,6 +832,11 @@ space is available. The value of @code{realloc} is the new address of the > block. If the block needs to be moved, @code{realloc} copies the old > contents. > > +The ISO C2x standard regards a zero value for @var{newsize} as undefined > +through N2464, so code that aims to be strictly compliant should avoid passing > +zero as @var{newsize} to @code{realloc}. @Theglibc{} behavior however is well > +defined below. There is no ISO C2x standard, and there will never be. I think it makes sense to add a sentence at the end of the section that behavior of realloc varies between implementations for zero arguments, without a standard reference, but mention the {Replace malloc} node (that is, an interposed malloc may behave differently). Thanks, Florian
On 2023-06-14 11:54, Florian Weimer wrote: > * Siddhesh Poyarekar: > >> Explicitly note that the realloc behaviour chosen in glibc is deliberate >> and well defined and suggest that strictly compliant code should avoid >> passing zero to realloc. >> >> Signed-off-by: Siddhesh Poyarekar <siddhesh@sourceware.org> >> --- >> manual/memory.texi | 5 +++++ >> 1 file changed, 5 insertions(+) >> >> diff --git a/manual/memory.texi b/manual/memory.texi >> index 5781a64f35..7697a10f72 100644 >> --- a/manual/memory.texi >> +++ b/manual/memory.texi >> @@ -832,6 +832,11 @@ space is available. The value of @code{realloc} is the new address of the >> block. If the block needs to be moved, @code{realloc} copies the old >> contents. >> >> +The ISO C2x standard regards a zero value for @var{newsize} as undefined >> +through N2464, so code that aims to be strictly compliant should avoid passing >> +zero as @var{newsize} to @code{realloc}. @Theglibc{} behavior however is well >> +defined below. > > There is no ISO C2x standard, and there will never be. We refer to it in arith.texi FWIW, which emboldened me to add that reference: manual/arith.texi:NaN is specified in C2x; the value was unspecified in TS 18661.) > I think it makes sense to add a sentence at the end of the section that > behavior of realloc varies between implementations for zero arguments, > without a standard reference, but mention the {Replace malloc} node > (that is, an interposed malloc may behave differently). Fair enough, v2 coming up. Thanks, Sid
diff --git a/manual/memory.texi b/manual/memory.texi index 5781a64f35..7697a10f72 100644 --- a/manual/memory.texi +++ b/manual/memory.texi @@ -832,6 +832,11 @@ space is available. The value of @code{realloc} is the new address of the block. If the block needs to be moved, @code{realloc} copies the old contents. +The ISO C2x standard regards a zero value for @var{newsize} as undefined +through N2464, so code that aims to be strictly compliant should avoid passing +zero as @var{newsize} to @code{realloc}. @Theglibc{} behavior however is well +defined below. + If you pass a null pointer for @var{ptr}, @code{realloc} behaves just like @samp{malloc (@var{newsize})}. Otherwise, if @var{newsize} is zero
Explicitly note that the realloc behaviour chosen in glibc is deliberate and well defined and suggest that strictly compliant code should avoid passing zero to realloc. Signed-off-by: Siddhesh Poyarekar <siddhesh@sourceware.org> --- manual/memory.texi | 5 +++++ 1 file changed, 5 insertions(+)