aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/net
diff options
context:
space:
mode:
authorJarek Poplawski <jarkao2@o2.pl>2007-07-26 08:44:01 -0400
committerJeff Garzik <jeff@garzik.org>2007-07-30 15:47:20 -0400
commit55b7b629b78cf82389baf604efcdd2b83b998ca7 (patch)
tree6c82f159c4bec8d27144c212bc5dc00f0f922898 /drivers/net
parent9351982b25ace7ee5ed82b6f4a7ea1151f31d267 (diff)
lib8390: comment on locking by Alan Cox
Additional explanation of problems with locking by Alan Cox. Signed-off-by: Jarek Poplawski <jarkao2@o2.pl> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Cc: Paul Gortmaker <p_gortmaker@yahoo.com> Cc: Jeff Garzik <jeff@garzik.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Diffstat (limited to 'drivers/net')
-rw-r--r--drivers/net/lib8390.c46
1 files changed, 46 insertions, 0 deletions
diff --git a/drivers/net/lib8390.c b/drivers/net/lib8390.c
index 721ee38d2241..c429a5002dd6 100644
--- a/drivers/net/lib8390.c
+++ b/drivers/net/lib8390.c
@@ -143,6 +143,52 @@ static void __NS8390_init(struct net_device *dev, int startp);
143 * annoying the transmit function is called bh atomic. That places 143 * annoying the transmit function is called bh atomic. That places
144 * restrictions on the user context callers as disable_irq won't save 144 * restrictions on the user context callers as disable_irq won't save
145 * them. 145 * them.
146 *
147 * Additional explanation of problems with locking by Alan Cox:
148 *
149 * "The author (me) didn't use spin_lock_irqsave because the slowness of the
150 * card means that approach caused horrible problems like losing serial data
151 * at 38400 baud on some chips. Rememeber many 8390 nics on PCI were ISA
152 * chips with FPGA front ends.
153 *
154 * Ok the logic behind the 8390 is very simple:
155 *
156 * Things to know
157 * - IRQ delivery is asynchronous to the PCI bus
158 * - Blocking the local CPU IRQ via spin locks was too slow
159 * - The chip has register windows needing locking work
160 *
161 * So the path was once (I say once as people appear to have changed it
162 * in the mean time and it now looks rather bogus if the changes to use
163 * disable_irq_nosync_irqsave are disabling the local IRQ)
164 *
165 *
166 * Take the page lock
167 * Mask the IRQ on chip
168 * Disable the IRQ (but not mask locally- someone seems to have
169 * broken this with the lock validator stuff)
170 * [This must be _nosync as the page lock may otherwise
171 * deadlock us]
172 * Drop the page lock and turn IRQs back on
173 *
174 * At this point an existing IRQ may still be running but we can't
175 * get a new one
176 *
177 * Take the lock (so we know the IRQ has terminated) but don't mask
178 * the IRQs on the processor
179 * Set irqlock [for debug]
180 *
181 * Transmit (slow as ****)
182 *
183 * re-enable the IRQ
184 *
185 *
186 * We have to use disable_irq because otherwise you will get delayed
187 * interrupts on the APIC bus deadlocking the transmit path.
188 *
189 * Quite hairy but the chip simply wasn't designed for SMP and you can't
190 * even ACK an interrupt without risking corrupting other parallel
191 * activities on the chip." [lkml, 25 Jul 2007]
146 */ 192 */
147 193
148 194