Hello, I finally just got my SOIC to DIP adapter for the 10-Bit FET Bus Switch, so I will get it together in the next day or two. I came up with this simple ppdev driver (attached) to test LED's connected to the data lines, But if all goes well it will be a good base for a Paraflasher->flashrom driver. Comments, suggestions?
Joseph Smith wrote:
I came up with this simple ppdev driver (attached) to test LED's connected to the data lines, But if all goes well it will be a good base for a Paraflasher->flashrom driver. Comments, suggestions?
Looks like it should work fine, but you can bet it will be pretty slow. :)
lmilk using direct port IO needs rougly 300 seconds to program a 256kbyte chip.
//Peter
On Wed, 17 Dec 2008 16:10:14 +0100, Peter Stuge peter@stuge.se wrote:
Joseph Smith wrote:
I came up with this simple ppdev driver (attached) to test LED's connected to the data lines, But if all goes well it will be a good base for a Paraflasher->flashrom driver. Comments, suggestions?
Looks like it should work fine, but you can bet it will be pretty slow. :)
lmilk using direct port IO needs rougly 300 seconds to program a 256kbyte chip.
Hmm, I like the direct IO approach also, but it would be the dirty way for a user-space driver. ppdev user-space driver is supposed to be the politically correct way. I'm stuck between a rock, can anyone else shed some light on this? Which way should we go on this?
Joseph Smith wrote:
Which way should we go on this?
It's not important, it can very easily be changed later. Just use something that works for you. /dev/parport is a nice safe Linux API.
I was just saying it is really slow so you know what to expect when testing. (I would guess about half the performance of lmilk, so maybe 512 bytes programmed per second.)
//Peter
On Wed, 17 Dec 2008 13:25:10 -0500, Joseph Smith joe@settoplinux.org wrote:
On Wed, 17 Dec 2008 16:10:14 +0100, Peter Stuge peter@stuge.se wrote:
Joseph Smith wrote:
I came up with this simple ppdev driver (attached) to test LED's connected to the data lines, But if all goes well it will be a good base for a Paraflasher->flashrom driver. Comments, suggestions?
Looks like it should work fine, but you can bet it will be pretty slow. :)
lmilk using direct port IO needs rougly 300 seconds to program a 256kbyte chip.
Hmm, I like the direct IO approach also, but it would be the dirty way
for
a user-space driver. ppdev user-space driver is supposed to be the politically correct way. I'm stuck between a rock, can anyone else shed some light on this? Which way should we go on this?
What about a combonation of the two. I could use the ppdev driver to control access to the parallel port and then just use direct IO reads and writes (outb and inb) directly to the port. I think if that will work, it will speed thing right up :-)
Hmm - do you receive my emails? Some have told me that my messages end up in their spamboxes.
Joseph Smith wrote:
What about a combonation of the two.
It's your code so you get to do whatever you want.
If you ask me, I strongly recommend depending on only _one_ interface.
//Peter
On Sat, 20 Dec 2008 22:09:34 +0100, Peter Stuge peter@stuge.se wrote:
Hmm - do you receive my emails? Some have told me that my messages end up in their spamboxes.
Yeh I get your emails.
Joseph Smith wrote:
What about a combonation of the two.
It's your code so you get to do whatever you want.
If you ask me, I strongly recommend depending on only _one_ interface.
Well, I like the way ppdev gains control over the port, but your right, we need speed and that is where direct IO access comes in.
Joseph Smith wrote:
Hmm - do you receive my emails? Some have told me that my messages end up in their spamboxes.
Yeh I get your emails.
Ok! Thanks for confirming.
What about a combonation of the two.
If you ask me, I strongly recommend depending on only _one_ interface.
Well, I like the way ppdev gains control over the port, but your right, we need speed and that is where direct IO access comes in.
My point was not that we need speed, but that ppdev will be so slow that you may not think that it is working correctly when starting to test with hardware.
Anyone who does need speed should better spend some money on a good standalone programmer.
If you are going to depend on direct IO access (with all that entails) then _additionally_ depending on a kernel driver that may not be available in the system makes little sense.
If you are going to depend on ppdev (with all that entails) then _additionally_ depending on the privileges required for direct IO makes little sense.
Either or. Pick one of them. They both have advantages and drawbacks.
Also, it's really poor form to start using a kernel driver only to bypass it with direct hardware access shortly thereafter. Boot DOS if you're into that kind of thing. :)
//Peter
-
Joseph Smith -
Peter Stuge