Обсуждение: [HACKERS] packetBuf size
Does anyone have any idea why the DATA variable in
PacketBuf is this MESSAGE_SIZE size in libpq?
#define MESSAGE_SIZE sizeof(StartupInfo)+5
^^^
Bruce, I think you might have commented on this before..I am not sure.
The total amount of data written into that variable is
database (64 bytes), username (32 bytes), options (64), execfile(64),
tty (64)...total is 288 bytes (MESSAGE_SIZE is 293). The last 5 bytes are
never used. Purify reports an uninitialized memory read when the packetBuf
is passed into sendto() function. Removing the +5 from the definition of
MESSAGE_SIZE removes this error, and I haven't noticed any problems after
running regression tests...
Comments?
=+=------------------------/\---------------------------------=+=
Igor Natanzon |**| E-mail: igor@sba.miami.edu
=+=------------------------\/---------------------------------=+=
------------------------------
> > Does anyone have any idea why the DATA variable in > PacketBuf is this MESSAGE_SIZE size in libpq? > #define MESSAGE_SIZE sizeof(StartupInfo)+5 > ^^^ > Bruce, I think you might have commented on this before..I am not sure. > The total amount of data written into that variable is > database (64 bytes), username (32 bytes), options (64), execfile(64), > tty (64)...total is 288 bytes (MESSAGE_SIZE is 293). The last 5 bytes are > never used. Purify reports an uninitialized memory read when the packetBuf > is passed into sendto() function. Removing the +5 from the definition of > MESSAGE_SIZE removes this error, and I haven't noticed any problems after > running regression tests... > > Comments? No one knows what that 5 is for. We are going to do some libpq communication changes in 6.2, so hopefully we will eliminate it then. I say remove the +5 as soon we start on 6.2, and see what happens. - -- Bruce Momjian maillist@candle.pha.pa.us ------------------------------
Yea...that sounds like a good idea. We will have ample time to find out if
it affects anything.
=+=------------------------/\---------------------------------=+=
Igor Natanzon |**| E-mail: igor@sba.miami.edu
=+=------------------------\/---------------------------------=+=
On Sat, 7 Jun 1997, Bruce Momjian wrote:
> >
> > Does anyone have any idea why the DATA variable in
> > PacketBuf is this MESSAGE_SIZE size in libpq?
> > #define MESSAGE_SIZE sizeof(StartupInfo)+5
> > ^^^
> > Bruce, I think you might have commented on this before..I am not sure.
> > The total amount of data written into that variable is
> > database (64 bytes), username (32 bytes), options (64), execfile(64),
> > tty (64)...total is 288 bytes (MESSAGE_SIZE is 293). The last 5 bytes are
> > never used. Purify reports an uninitialized memory read when the packetBuf
> > is passed into sendto() function. Removing the +5 from the definition of
> > MESSAGE_SIZE removes this error, and I haven't noticed any problems after
> > running regression tests...
> >
> > Comments?
>
> No one knows what that 5 is for. We are going to do some libpq
> communication changes in 6.2, so hopefully we will eliminate it then.
>
> I say remove the +5 as soon we start on 6.2, and see what happens.
>
>
>
> --
> Bruce Momjian
> maillist@candle.pha.pa.us
>
------------------------------