Revert "yenta free_irq on suspend"
authorLinus Torvalds <torvalds@g5.osdl.org>
Sat, 30 Jul 2005 20:41:56 +0000 (13:41 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Sat, 30 Jul 2005 20:41:56 +0000 (13:41 -0700)
ACPI is wrong.  Devices should not release their IRQ's on suspend and
re-aquire them on resume.  ACPI should just re-init the IRQ controller
instead of breaking most drivers very subtly.

Breakage reported by Hugh Dickins <hugh@veritas.com>

Undo: d8c4b4195c7d664baf296818bf756775149232d3

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
drivers/pcmcia/yenta_socket.c

index 744e469a9eda6cca3e0e45457706442e99f3f687..6837491f021c0b813aee8681cc73eb33dd0a21cd 100644 (file)
@@ -1107,8 +1107,6 @@ static int yenta_dev_suspend (struct pci_dev *dev, pm_message_t state)
                pci_read_config_dword(dev, 17*4, &socket->saved_state[1]);
                pci_disable_device(dev);
 
-               free_irq(dev->irq, socket);
-
                /*
                 * Some laptops (IBM T22) do not like us putting the Cardbus
                 * bridge into D3.  At a guess, some other laptop will
@@ -1134,13 +1132,6 @@ static int yenta_dev_resume (struct pci_dev *dev)
                pci_enable_device(dev);
                pci_set_master(dev);
 
-               if (socket->cb_irq)
-                       if (request_irq(socket->cb_irq, yenta_interrupt,
-                                       SA_SHIRQ, "yenta", socket)) {
-                               printk(KERN_WARNING "Yenta: request_irq() failed on resume!\n");
-                               socket->cb_irq = 0;
-                       }
-
                if (socket->type && socket->type->restore_state)
                        socket->type->restore_state(socket);
        }