-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all i2c hackers, (I've added Ben Dooks as the author of relevant code)
> I found the reason why the i2c is on the phone so slow. > The phone uses the driver i2c-s3c2410.c > And after each acces to i2c there is the function msleep(1). > Which needs on the openmoko about 10ms. Yes, I can confirm this. I'm connecting some prototype pointing device that needs to access several i2c registers on each readout of current position and having some 100ms delay here is really bad (both the delay as well as limited readout frequency). > I made a little patch for that and now i2c is quite fast. > Maybe that should help someone. One I2C-Access needs now about ~0.6ms > and before it was 10ms. > I attached the patch. With this patch I sometimes can't even boot my neo (gta02a6) :( but yes, when it manages to fully boot it works much better. However, those udelays() are most probably not really good for the rest of the system.. Is there anything that could be done to fix this? this limits i2c usability a lot :( Regards, Vladimir > > Kind regards, > Matthias Katzengruber > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iJwEAQECAAYFAkk+lEcACgkQutgEj9ZLuzR91AP+L1ur0E4blVbtZa7uoe489chC aEwnh8yCP6MZ1THtu4lbHRP8/ppoF4mphrqoXuuKxA2gcMF0Mk3abCeLtMhYboij YqI/VlC4BeGG82kj8Ms9iuPOJWsW2YBWNmG7Eaayf4A1fPAZLKnOR9/28Am8DwdI xYD8WFRr9IS3bBgo/ew= =EIS5 -----END PGP SIGNATURE-----
