On Sat, Mar 16, 2013 at 1:00 PM, Henrik Rydberg <rydberg@xxxxxxxxxxx> wrote: >> Userspace can initiate the firmware update procedure by copying cyapa.bin >> to /lib/firmware, and then writing anything to the device sysfs attribute: >> /sys/bus/i2c/devices/7-0067/update_fw > > Why do you need to trigger this from userland via sysfs? > It's a confluence of factors that drives the decision trigger from userland. - As a part of the cold-boot probe steps, the trackpad's bootloader will load an operational firmware from its flash memory. In the average case, the trackpad driver does not need to update the firmware because it's current. (the non-average cases are cases where on device firmware is corrupted or out of date) - The cypress firmware payloads we are working with here do not contain an indication of firmware version in the payload itself, so the driver can't tell at probe time whether what it would get if it request_firmware is newer, older, or the same version as what's already loaded on the trackpad. - On a 400kHz i2c bus, it takes 13 seconds to send the 31K payload to the device and to wait for it to jump into the operational mode firmware. This is too long to block probe every time. - Because it takes so long to do a firmware update, our user space would like to put up a warning message anyhow to tell the user that the touchpad is being updated, and not to disturb the system. >> + if (sysfs_create_group(&client->dev.kobj, &cyapa_sysfs_group)) >> + dev_warn(dev, "error creating sysfs entries.\n"); >> + > > Would it not be neat to invoke the firmware update automatically instead? The one situation where I could see an automatic firmware update being useful would be cases where the driver, during probe, discovered that the trackpad is in Bootloader mode because no valid firmware was found (a case of corruption of some kind due to the system powering down during a previous fw update, or example). In that case, the driver can request firmware and update it without user space having to trigger it later. Another driver in input does it the same way. atmel_mxt_ts.c, which our team is actively working on as well, also has an update_fw sysfs for a lot of the same reasons. Thanks, -- Benson Leung Software Engineer, Chrom* OS bleung@xxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
