13 messages in org.freebsd.freebsd-archnewbus flaw| From | Sent On | Attachments |
|---|---|---|
| Dag-Erling Smørgrav | May 11, 2004 7:39 am | |
| Doug Rabson | May 11, 2004 8:08 am | |
| Poul-Henning Kamp | May 11, 2004 8:12 am | |
| John-Mark Gurney | May 11, 2004 6:02 pm | |
| Dag-Erling Smørgrav | May 12, 2004 6:27 am | |
| John-Mark Gurney | May 12, 2004 10:41 am | |
| Dag-Erling Smørgrav | May 12, 2004 10:48 am | |
| John-Mark Gurney | May 12, 2004 10:53 am | |
| Dag-Erling Smørgrav | May 12, 2004 1:30 pm | |
| Doug Rabson | May 13, 2004 1:27 am | |
| Dag-Erling Smørgrav | May 13, 2004 3:46 am | |
| Doug Rabson | May 13, 2004 5:34 am | |
| M. Warner Losh | May 13, 2004 9:06 am |
| Subject: | newbus flaw | |
|---|---|---|
| From: | John-Mark Gurney (gurn...@efn.org) | |
| Date: | May 11, 2004 6:02:34 pm | |
| List: | org.freebsd.freebsd-arch | |
Dag-Erling Sm?rgrav wrote this message on Tue, May 11, 2004 at 16:39 +0200:
When the driver is unloaded, the device is detached, but it remains on the bus's list of child devices. The next time the module is loaded, its DEVICE_IDENTIFY method is called again, and incorrectly adds a second child device to the bus, because it does not know that one already exists.
There is no way for DEVICE_IDENTIFY to check if a matching child already exists on the bus, or for the module's event handler to unlist the child when unloading.
As someone else points out, this is already a known case and was discusses on -arch a while back.
You are incorrect in assuming you can't find out if another child already exists.. Usually this is a problem of properly allocating resources so that you know the other child exists. Since you are using identify, you already don't have a "self describing" bus, which means that you have to either use hints, or another method to make sure that your device doesn't already exist.
For example, in my adv717xa driver, part of the zoran driver, I do the following in my _probe routine to prevent probing and attaching to the same chip. /* Make sure this isn't already attached */ if (sibs == NULL) device_get_children(iicbus, &sibs, &sibcnt); for (j = 0; j < sibcnt; j++) { if (device_get_state(sibs[j]) >= DS_ATTACHED && device_get_driver(sibs[j]) == drv && ((struct adv717xa_softc *) device_get_softc(sibs[j]))->adv_slvaddr == chips[i].addr) { device_printf(dev, "already attached at %s\n", device_get_nameunit(sibs[j])); break; } }
This should not be a problem because the i2c bus should either do proper resource allocation for the id, which when I try to probe, I allocate the chip id, and that fails, preventing it's creation...
-- John-Mark Gurney Voice: +1 415 225 5579
"All that I will do, has been done, All that I have, has not."