Hi- I'm trying to determine if what I'd like to do will cause problems. We have a PCI device where we can look at a byte of it's memory to determine what slot our linux based management board is in (non-prefetchable PCI memory). The device also does a bunch of other things beyond the scope of what I'd like to do. We have a driver for the device that allows a user space process to open it up, and mmap the memory area for the PCI device so the user space code can see what slot we're in, and look at whatever else. BTW the "slot" I'm talking about has nothing to do with PCI slots -- it's the slot the entire system is in. The whole "linux box" is plugged into a slot in a chassis. I have another driver and from within that driver, I'd like to be able to see what slot I'm in. So I've got code that works where it finds the PCI device address (it's something like 0xfedfC000) from the PCI config info and uses ioremap() to map the device address to kernel virtual memory and I go look at the slot. Then I iounmap() and go onward doing my driver's functionality (which doesn't use that PCI device anymore). My question is, if a user space process has opened the other driver and mmapped() the same region that my driver ioremaps for a moment, will there be a conflict? i.e. will the ioremap I'm doing affect the already mmapped area? I'm thinking the kernel virtual memory's mapping to the PCI device should be independent of the userspace mmapped mapping to the PCI device and my driver and the userspace code using the other driver can read the slot info w/o problems if timing is such that the ioremap in my driver is done at a point where the user space process has the region mmaped through the other driver. Is this is a safe assumption? A quick breeze through do_mmap and ioremap look like they're independent but I'm not sure as I obviously don't have complete comprehension of the memory managment stuff. It took me a while to realize I had to ioremap that high address to be able to look at the PCI memory w/o causing a kernel seg fault (unable to handle paging request). ioremap fixed that for me nicely. But I'm wondering if I should be concerned about coordination between my driver and the "other" driver. Thanks for any info, -Eric Malkowski -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ IRC Channel: irc.openprojects.net / #kernelnewbies Web Page: http://www.kernelnewbies.org/
