Re: [PATCH 1/2] sideband.c: Use xmalloc() instead of variable-sized arrays.

Nicolas Pitre [off-list ref] writes:

On Tue, 8 Jan 2008, Junio C Hamano wrote:

quoted

Johannes Sixt [off-list ref] writes:

quoted

From: Johannes Sixt <redacted>

How come we got along with this not very portable construct for so long?
Probably because the array sizes were computed from the results of
strlen() of string constants. Anyway, a follow-up patch will make the
lengths really non-constant.

Signed-off-by: Johannes Sixt <redacted>
---
 sideband.c |   14 ++++++++++++--
 1 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/sideband.c b/sideband.c
index 756bbc2..513d7b3 100644
--- a/sideband.c
+++ b/sideband.c

@@ -19,7 +19,10 @@ int recv_sideband(const char *me, int in_stream, int out, int err)
 {
 	unsigned pf = strlen(PREFIX);
 	unsigned sf = strlen(SUFFIX);
-	char buf[pf + LARGE_PACKET_MAX + sf + 1];
+	char *buf, *save;
+
+	save = xmalloc(sf);
+	buf = xmalloc(pf + LARGE_PACKET_MAX + sf + 1);

I have to wonder if the malloc() overhead is small enough
compared to the network bandwidth to make a two malloc-free
pairs per packet a non-issue...

Eeek.  Overhead might be insignificant, but it still doesn't feel right.

What about using alloca() instead?  This is not like if we are doing 
funky things with the allocated memory anyway.

That's double Eek as I recall AIX is not dead.

How about using a constant large enough slop?  It is not like
PREFIX and SUFFIX are different vastly between calls.

On Tue, 8 Jan 2008, Junio C Hamano wrote:

How about using a constant large enough slop?  It is not like
PREFIX and SUFFIX are different vastly between calls.

The size of the buffer is certainly bounded since the prefix length is 
constant, the buffer size is constant, and both possible suffixes are 
also constants.


Nicolas