PATCH: IPMI driver
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 7 lines
ChangeSet 1.980, 2003/01/08 22:23:15-02:00, [email protected]
[PATCH] PATCH: IPMI driver
This has been in -ac for a short while. Linus accepted and merged the same IPMI support into 2.5.54 so now it can move into 2.4 IMHO
Quoted 13 lines
Documentation/IPMI.txt | 341 ++++++ drivers/char/Makefile | 5 drivers/char/ipmi/Makefile | 20 drivers/char/ipmi/ipmi_devintf.c | 532 ++++++++++ drivers/char/ipmi/ipmi_kcs_intf.c | 1235 ++++++++++++++++++++++++ drivers/char/ipmi/ipmi_kcs_sm.c | 467 +++++++++ drivers/char/ipmi/ipmi_kcs_sm.h | 70 + drivers/char/ipmi/ipmi_msghandler.c | 1811 ++++++++++++++++++++++++++++++++++++ drivers/char/ipmi/ipmi_watchdog.c | 971 +++++++++++++++++++ include/linux/ipmi.h | 516 ++++++++++ include/linux/ipmi_msgdefs.h | 58 + include/linux/ipmi_smi.h | 144 ++ 12 files changed, 6170 insertions(+)
Either I'm blind, or none of those files exist in Linus' tree looking at current bitkeeper snapshot.
Dave
Signature · 2 lines
| Dave Jones. http://www.codemonkey.org.uk | SuSE Labs
Quoted 3 lines · Dave Jones
Either I'm blind, or none of those files exist in Linus' tree looking at current bitkeeper snapshot.
Arghhh I was told Linus accepted it, and my tree indexer found "IPMI" so decided it was present too. (Only the i2c definitions apparently). Oh well, it should be in 2.5
Quoted 2 lines · Alan Cox
Arghhh I was told Linus accepted it, and my tree indexer found "IPMI" so decided it was present too. (Only the i2c definitions apparently).
Shouldn't cause any problems in 2.4 anyways should it ? After all, its 'just another driver'.
Quoted 1 line
Oh well, it should be in 2.5
Added to the queue of bits from the 2.4 changesets list that I'm intending to push to Linus soon.
Dave
Signature · 2 lines
| Dave Jones. http://www.codemonkey.org.uk | SuSE Labs
Quoted 5 lines · Dave Jones
Quoted 2 lines · Alan Cox
Arghhh I was told Linus accepted it, and my tree indexer found "IPMI" so decided it was present too. (Only the i2c definitions apparently).
Quoted 7 lines
Shouldn't cause any problems in 2.4 anyways should it ? After all, its 'just another driver'.
Quoted 1 line
Oh well, it should be in 2.5
Added to the queue of bits from the 2.4 changesets list that I'm intending to push to Linus soon.
Pull the 2.5 port from openipmi.sourceforge.net saves you doing the port yourself.
Quoted 13 lines · Alan Cox
Quoted 10 lines · Dave Jones
Quoted 2 lines · Alan Cox
Arghhh I was told Linus accepted it, and my tree indexer found "IPMI" so decided it was present too. (Only the i2c definitions apparently).
Shouldn't cause any problems in 2.4 anyways should it ? After all, its 'just another driver'.
Quoted 1 line
Oh well, it should be in 2.5
Quoted 8 lines
Quoted 4 lines
Added to the queue of bits from the 2.4 changesets list that I'm intending to push to Linus soon.
Pull the 2.5 port from openipmi.sourceforge.net saves you doing the port yourself.
Definately pull the 2.5 port from there, as there are some differences between the 2.4 and 2.5 versions.
-Corey
Quoted 5 lines · Corey Minyard
Quoted 3 lines
Pull the 2.5 port from openipmi.sourceforge.net saves you doing the port yourself.
Definately pull the 2.5 port from there, as there are some differences between the 2.4 and 2.5 versions.
I had a quick skim through the patch.
Is the handling of jiffies wraps done correctly ? It doesn't look like it...
time_diff = ((jiffies_now - kcs_info->last_timeout_jiffies)
Dave
Signature · 1 line
| Dave Jones. http://www.codemonkey.org.uk
Quoted 1 line · Dave Jones
time_diff= ((jiffies_now-kcs_info->last_timeout_jiffies)
Thats valid for unsigned maths
0x00000001 - 0xFFFFFFFF = 0x00000002
Alan
Quoted 5 lines · Alan Cox
Quoted 1 line · Dave Jones
time_diff= ((jiffies_now-kcs_info->last_timeout_jiffies)Thats valid for unsigned maths
0x00000001 - 0xFFFFFFFF = 0x00000002
Doh, I've made this mistake before.. Thanks for clarifying.
Dave
Signature · 1 line
| Dave Jones. http://www.codemonkey.org.uk
Quoted 5 lines · Alan Cox
Quoted 1 line · Dave Jones
time_diff= ((jiffies_now-kcs_info->last_timeout_jiffies)Thats valid for unsigned maths
0x00000001 - 0xFFFFFFFF = 0x00000002
Just as a general note (not to Alan), this often appears in ethernet drivers, in their RX and TX producer/consumer ring counters... so don't be surprised if you see this logic elsewhere in the kernel, too.
Jeff
Quoted 7 lines · Dave Jones
Quoted 4 lines · Corey Minyard
Quoted 3 lines
Pull the 2.5 port from openipmi.sourceforge.net saves you doing the port yourself.
Definately pull the 2.5 port from there, as there are some differences between the 2.4 and 2.5 versions.
Quoted 8 lines
I had a quick skim through the patch.
Is the handling of jiffies wraps done correctly ? It doesn't look like it...
time_diff= ((jiffies_now-kcs_info->last_timeout_jiffies)Dave
I don't understand why that wouldn't work. Those are both unsigned long values. Assuming twos complement, the time difference could be correct, even in a wraparound case (unless a very large number of jiffies has transpired, but that will never be the case here).
Am I missing something here?
-Corey