In message <[EMAIL PROTECTED]>, Julian Elischer writes:
>1/ I think the removal of IPSEC from LINT is by mistake in this patch
yes, that is a mistake.
>2/ The change of separating buffer management from IO management
>is long overdue and the introduction of b_iocmd is a good first step
>for this. The selection of values is in my mind arguable either way.
>
>As the field is an 'op-code' I would have chosen the values to be:
>#define BIO_NOP 0
>#define BIO_READ 1
>#define BIO_WRITE 2
>#define BIO_DELETE 3
I decided to use binary values, because there is a simple arithmetic
test for "exactly one bit set" (See BUF_STRATEGY). I wanted to
catch as many bogus cases of B_WRITE zeroness being abused as possible,
ie, people initializing b_flags = 0 knowing that would give them
a write.
>4/ does this produce compatibility problems with any 3rd party code?
This is 5.0-CURRENT, they will have to adopt to much worse things
before it becomes 5.0-RELEASE.
--
Poul-Henning Kamp FreeBSD coreteam member
[EMAIL PROTECTED] "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
