Messages in this thread | | | Date | Mon, 8 Jun 1998 21:52:29 +0200 (MET DST) | From | Gerard Roudier <> | Subject | Re: piss poor change in ncr53c8xx/linux-2.1.104 |
| |
On Sun, 7 Jun 1998, Pavel Machek wrote:
> Hi! > > > FOR YOUR INFORMATION, THIS DRIVER IS STILL MAINTAINED. > > No need to be rude, that person did not have any bad intention. (I was > not that person, FYI). > > > FYI, this change will not be incorporated in my personal driver version, > > which is currently 2.6n and will become 3.0 soon with a patch against > > 2.0.34 followed by the announce that I stop maintaining ncr53c8xx 2.5 > > driver series. > > Ook, what about following patch? (I consider having DELAY() function > when there are generic mdelay() and udelay() functions available bad > thing.)
[ Patch that waste net bandwitch deleted ]
I just have had a look into the mdelay() stuff and I noticed the following:
1) There is LOTS of place where mdelay() is now used in the kernel. With fast modern CPU, it is just wasting things. 2) Tt is implemented from macros using udelay. So it will waste more memory than a simple function without being more accurate. 3) The only place where the change has been just crap is in the ncr53c8xx driver.
IMO, this function should be defined as follow:
#define mdelay(n) \ panic("Better to die than wasting memory and CPU stupidly\")
I am not going to use this useless thing that is designed to waste memory and CPU at the same time.
FYI, the ncr driver has never used more that 1000 as argument to udelay(). It is not my fault if lots of other kernel modules did'nt use udelay() as it was required.
Please, stop wasting my time with so obvious stuff as mdelay().
THAT'S ALL.
Gerard.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |