Re: WARNING: that patch from Ted...

Colin Plumb (colin@nyx.net)
Sun, 27 Apr 97 20:30:09 MDT

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Victor Yodaiken: "Re: RT-Linux and SMP"
Previous message: Richard B. Johnson: "DEATH of SMP"
Maybe in reply to: Colin Plumb: "WARNING: that patch from Ted..."

Er, um, I'm *really* sorry to point out bugs twice in a row, but when
merging our patches, you made a mistake. To be precise, the
+ seq = (halfMD4Transform(secret+8, secret) &
+ ((1<<HASH_BITS)-1)) + (count << HASH_BITS);
part. I left the << HASH_BITS out because I changed the code so count
was a __u32, pre-shifted. Thus, you don't *want* the count, this
shift makes the result always 0, and the ISN will thus occasionally step
backwards, a not-so-wonderful thing.

Sorry, I should have made that little speed optimization (this whole patch
is a speed optimization, so it seemed apropos) clearer when I sent it to
you.

-- 
	-Colin

Next message: Victor Yodaiken: "Re: RT-Linux and SMP"
Previous message: Richard B. Johnson: "DEATH of SMP"
Maybe in reply to: Colin Plumb: "WARNING: that patch from Ted..."