[PATCH] PCMCIA updates

Russell King (rmk@arm.linux.org.uk)
Tue, 22 Apr 2003 01:19:20 +0100


A request for Linus to pull a set of PCMCIA changes has been made. You
can get the patch from:

http://patches.arm.linux.org.uk/pcmcia-20030421.diff

Much of this stuff has been aired on the linux-pcmcia list (iirc care of
lists.infradead.org)

This will update the following files:

drivers/net/wireless/orinoco_cs.c | 1
drivers/pcmcia/cs.c | 827 +++++++++++++++++++++++++++-----------
drivers/pcmcia/cs_internal.h | 18
drivers/pcmcia/rsrc_mgr.c | 20
include/pcmcia/bus_ops.h | 2
include/pcmcia/driver_ops.h | 2
include/pcmcia/ds.h | 2
7 files changed, 617 insertions, 255 deletions

through these ChangeSets:

<proski@org.rmk.(none)> (03/04/21 1.1122)
[PCMCIA] validate_mem uses uninitialized memory.

Patch from Pavel Roskin.

If I compile a recent 2.5.x kernel without CONFIG_ISA defined, I get
an oops in validate_mem(). Stack trace contains 0x6b6b6b6 - a clear
sign that freed memory is being accessed.

It's the second validate_mem() in drivers/pcmcia/rsrc_mgr.c - the one
used when CONFIG_PCMCIA_PROBE is not defined. It turns out the memory
is freed in do_mem_probe() when it's called from validate_mem().

The solution is to use the same trick as in the first validate_mem().
This problem is quite serious and it's not specific to the plx9052
driver. I see it with yenta_socket as well.

<linux@de.rmk.(none)> (03/04/19 1.1121)
[PCMCIA] remove unused files [Christoph Hellwig]

From Christoph Hellwig

There's no need to keep the stubs around.

<linux@de.rmk.(none)> (03/04/19 1.1120)
[PCMCIA] Fix compilation of cardmgr [Pavel Roskin]

From Pavel Roskin

ds.h should not be including linux/device.h when compiling userspace
code.

<rmk@flint.arm.linux.org.uk> (03/04/19 1.1119)
[PCMCIA] Bring in PCMCIA work queue and state machine.

This cset fixes the boot-time deadlock which occurs when a Cardbus
card is inserted. The deadlock occurs because:

- the ds module is inserted
- ds registers a driver model interface for pcmcia socket drivers,
which takes the global devclass_sem.
- ds causes the pcmcia core to evaluate the status of the sockets,
and perform "card insertion" processing if cards are present.
- this processing detects a cardbus card, and calls the cardbus code
to scan pci devices, and add them to the device tree.
- each device gets passed to the device model's class layer, which
tries to take devclass_sem. However, we locked it while
initialising the ds module -> deadlock.

We fix this by making the card insertion/removal asynchronous to the
initialisation thread by using a work queue. However, since we will
be calling drivers, which may want to use workqueue services, we
need to use a separate workqueue.

We also bring in a state machine to handle the card insertion,
removal, suspend and reset code. This ensures that two operations
can not run against a single socket at the same time.

Unfortunately, the workqueue model has an annoying feature which
causes it to we spawn N threads, one for each CPU in the system,
when we create our work queue. This is rather overkill.

-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/