I will test soon I will give you the results.
Andre ha scritto: > > > You're right. At this point I focussed on the basic read/write and > compatibility stuff. Functions like that can be integrated, if > interest exist. > > Right now, I would like to know whether people, who also had issues > with the existing driver, find this version an improvement. > > --- In [email protected] <mailto:foxboard%40yahoogroups.com>, > d191264it <[EMAIL PROTECTED]> wrote: >> >> I worked also on that driver in the past and I found that tools like >> e2fsck was not working properly due to some missing calls. >> So I added functions like this: >> /******************************************************************* > ******/ >> static int mmc_getgeo(struct block_device *bdev, struct hd_geometry > *geo) >> { >> VOLUME_INFO info; >> MMC_get_volume_info(&info); >> >> geo->heads = 4; >> geo->sectors = 61; >> geo->cylinders = (info.size / (512*4*61)); >> return 0; >> } >> static int mmc_ioctl(struct inode * inode, struct file * file, >> unsigned int cmd, unsigned long arg) >> { >> VOLUME_INFO info; >> struct hd_geometry __user *loc = (struct hd_geometry __user > *) arg; >> struct hd_geometry g; >> >> if (cmd != HDIO_GETGEO) >> return -EINVAL; >> if (!loc) >> return -EINVAL; >> >> MMC_get_volume_info(&info); >> >> g.heads = 4; >> g.sectors = 61; >> g.cylinders = (info.size / (512*4*61)); >> g.start = get_start_sect(inode->i_bdev); >> return copy_to_user(loc, &g, sizeof g) ? -EFAULT : 0; >> } >> >> static struct block_device_operations bdops = { >> .open = mmc_do_open, >> .release = mmc_release, >> .media_changed = mmc_has_media_changed, >> .revalidate_disk = mmc_revalidate, >> .ioctl = mmc_ioctl, >> .getgeo = mmc_getgeo, >> .owner = THIS_MODULE, >> }; >> /******************************************************************* > ******/ >> >> should be fine to have it also in your driver >> >> >> >> >> Andre ha scritto: >> > >> > >> > Performance was not the primary reason for developing this driver. >> > I had some issues using large cards which were not recognised, > also >> > swapping cards frequently made the Fox unstable. >> > >> > Both drivers use the same type software spi, so interface speed > is in >> > the same range. I did implement stream mode reading/writing which >> > makes better use of the card's internal buffers. This typically > makes >> > larger transfers more effective. >> > For full effect, this requires the card to be mounted without any >> > option. (constantly writing to the fat doesn't sounds like a good >> > idea anyway) >> > >> > This way I get about 100 - 200 kiB/s, (regular SD cards) using > Debian >> > ftp client to the Fox ftp server. New or erased cards may perform >> > somewhat better. >> > >> > --- In [email protected] <mailto:foxboard%40yahoogroups.com> > <mailto:foxboard% > 40yahoogroups.com>, >> > John Crispin <john@> wrote: >> >> >> >> what is the performance like in comparison to the existing > driver ? >> >> >> >> >> >> Quoting afmdsgn <afmdsgn@>: >> >> >> >> > Hi All, >> >> > >> >> > As I had some issues with the MMC driver in the SDK, I > developed >> > a new >> >> > version that also supports the 2 GB SD and 4 GB HCSD. >> >> > >> >> > People interested can have a look at >> >> > http://www.afmdesign.net/proj/mmc/mmc.html > <http://www.afmdesign.net/proj/mmc/mmc.html> >> > <http://www.afmdesign.net/proj/mmc/mmc.html > <http://www.afmdesign.net/proj/mmc/mmc.html>> >> >> > >> >> > Please discuss any issues in this group. >> >> > >> >> > Greetings Andre. >> >> > >> >> > >> >> >> > >> > >> Chiacchiera con i tuoi amici in tempo reale! >> http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com > <http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com> >> > > Chiacchiera con i tuoi amici in tempo reale! http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com
