Skip to contentSkip to conversation

: Message Scheme, McDonald Patches

PATCH In review Gerrit Renker 1 participant 1 message Nov 20, 2006

lore
Share & tools
lore
#0 lore raw

cover : Message Scheme, McDonald Patches

This set contains
* an update of my patches from Friday
* Ian's patches

1) Reason for update
--------------------
The whole point of the message-macro scheme was to allow an easier migration later on to the well-known counterparts BUG(), WARN_ON(), BUG_ON(). These however issue a panic which stops the machine. Now, patch xx/7 sent in the last batch replaced BUG/BUG_ON with the DCCP counterparts. This is *not* a good idea, since these tests are in places where continuing would mean entering an undefined state. I therefore retract the said patch and have updated the others as pointed out below.

2) What changes:
----------------

	* Patch 1      (Ian's tx_qlen patch) remains as 9z.
	* Patch 2      (message scheme) will be re-sent as one patch (10a) instead of 7 single ones
	* Patch 3      (socket annotations) will be re-sent as second patch (10b)
	* Patch 4..6   are Ian's patches from Sunday and are not re-sent, but simply and shamelessly 

stacked them on top of all other patches as 11a .. 11c

3) Detailed breakdown of changes with regard to the earlier cset
----------------------------------------------------------------
Patch 2: Simplifications for ccid3.c
- remains as before

Patch 3: - CHANGED:
- removed superfluous DCCP_WARN_ON (no point using a DCCP_XX wrapper)
Patch 4: - Migration towards DCCP_BUG{,_ON}()
- RETRACTED: due to reasons above
Patch 5: - Migration towards DCCP_WARN_ON();
- RETRACTED: DCCP_BUG_ON is the same as WARN_ON, so not necessary
- remainder of Patch 4/5; replaces unconditional BUG_ON(1) / WARN_ON(1)

Patch 6:      - remains as before
Patch 7:      - remains as before

Patch 8: - done the FIXMes

-- Gerrit