- Do not leak memory while flushing rules.
Noticed by: yar
- Remove useless NULL pointer check: we are using M_WAITOK flag
for memory allocation.
Submitted by: Andrey Elsukov <bu7cher at yandex dot ru>
The high points for next RELENG_6 release are:
+ Fix 2322/6322 firmware loading- this card works a lot better with
loaded f/w.
+ Fix a considerable number of 2KLogin issues.
+ Clean all mailbox structures prior to use.
+ Remove some obsolete code within defines that haven't been
enabled in years.
+ Make Loop ID 125 a legal hard address (which it is).
+ Force HARD ADDRESS acquisition for all f/w revisions > 2.2.5
+ Use RIO_16BIT for 2200 cards >= 1.17.0 f/w, and enable
fast posting.
+ Honor ZIO nvram settings for 23XX cards.
+ Add an ISP_CRASHED state so that if we get a DMA error and the f/w
crashes we can unstall any mailbox commands that might have been
waiting.
+ Catch RISC PAUSED errors on interrupts.
+ Plus many, many more!
* rev. 1.21:
Don't push a map on a master server which will fail anyway.
PR: bin/72881
* rev. 1.22:
Remove unsafe use of asynchronous I/O and improve error handling.
PR: bin/102143
VIA C7. Although padlock(4) supports some of the crypto functionality
on the VIA C7, I mistakenly thought that its functionality on RELENG_6
was equivalent to that on HEAD, which isn't true, at least not yet.
Pointed out by: pjd (about a month ago)
Update two existing errata entries with pointer to EN-06:02.net (which
fixes them); add a new entry for the third item fixed by EN-06:02.net
(a local kernel panic not previously mentioned).
Remove merge plans for snapshot/deadlock fixes; due to the effort
involved, the likely fix for 6.1 users will be to update to 6.2.
- Use our standard section 4 SYNOPSIS layout.
- Remove the references to the README. All information the references
refer to are actually more up to date in our manpages.
Follow the lead of mxge(4) and build the module for ixgb(4).
Originally it wasn't enabled since the hardware wasn't commonplace,
but as 10GE hardware is becoming more widely used, building the module
by default should be beneficial.
Approved by: rwatson (mentor)
Define mtu as u_int16_t not as int. This should fix problem with rfcomm
on sparc64.
Reported by: Andrew Belashov <bel at orel dot ru>
Tested by: Andrew Belashov <bel at orel dot ru>
