[patch] sound: fix bootup crash in snd_gus_interrupt()

From: Ingo Molnar
Date: Tue Oct 16 2007 - 01:58:32 EST

Next message: Anton Arapov: "Re: [PATCH] ipv4: kernel panic when only one unsecured port available"
Previous message: Stefan Richter: "Re: What still uses the block layer?"
Next in thread: Takashi Iwai: "Re: [patch] sound: fix bootup crash in snd_gus_interrupt()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Subject: sound: fix bootup crash in snd_gus_interrupt()
From: Ingo Molnar <mingo@xxxxxxx>

when simulating a storm of fake GUS interrupts (without actually owning
this venerable piece of ISA hardware) the driver falls over (crashes) in
two ways:

1) spinlocks being initialized too late:

INFO: trying to register non-static key.
the code is fine but needs lockdep annotation.
turning off the locking correctness validator.
[<401058ca>] show_trace_log_lvl+0x1a/0x30
[<401064b2>] show_trace+0x12/0x20
[<401064d6>] dump_stack+0x16/0x20
[<4014a72b>] __lock_acquire+0xcfb/0x1030
[<4014aac0>] lock_acquire+0x60/0x80
[<40721a68>] _spin_lock_irqsave+0x38/0x50
[<4058fc12>] snd_gf1_i_look8+0x22/0x60
[<405906fe>] snd_gus_interrupt+0x13e/0x270
[<401548e8>] handle_IRQ_event+0x28/0x60
[<40155cc1>] handle_fasteoi_irq+0x71/0xe0
[<40107238>] do_IRQ+0x48/0xa0
[<401051fe>] common_interrupt+0x2e/0x40
[<40156822>] register_handler_proc+0x92/0xf0
[<401550c2>] setup_irq+0xe2/0x190
[<40155224>] request_irq+0xb4/0xd0
[<4058f524>] snd_gus_create+0x124/0x3c0
[<40aa4087>] snd_gusclassic_probe+0x2a7/0x4b0
[<403f5eff>] isa_bus_probe+0x1f/0x30
[<403f1944>] driver_probe_device+0x84/0x190
[<403f1a58>] __device_attach+0x8/0x10
[<403f0e63>] bus_for_each_drv+0x53/0x80
[<403f1b1b>] device_attach+0x8b/0x90
[<403f0dd8>] bus_attach_device+0x48/0x80
[<403efdbd>] device_add+0x45d/0x5a0
[<403eff12>] device_register+0x12/0x20
[<403f60c3>] isa_register_driver+0xb3/0x140
[<40aa3dd2>] alsa_card_gusclassic_init+0x12/0x20
[<40a665c3>] kernel_init+0x133/0x310
[<401054a7>] kernel_thread_helper+0x7/0x10
=======================

2) callback functions not being filled in yet:

BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000
printing eip:
00000000
*pde = 00000000
Oops: 0000 [#1]
SMP DEBUG_PAGEALLOC
CPU: 0
EIP: 0060:[<00000000>] Not tainted VLI
EFLAGS: 00010002 (2.6.23 #37)
EIP is at 0x0
eax: 7fe94000 ebx: 7fe94000 ecx: 00000000 edx: 00000226
esi: 00000000 edi: 00000005 ebp: 7ff87c28 esp: 7ff87bf4
ds: 007b es: 007b fs: 00d8 gs: 0000 ss: 0068
Process swapper (pid: 1, ti=7ff86000 task=7ff84000 task.ti=7ff86000)
Stack: 40590683 408424a9 408db87c 00000029 40787406 00000064 00000046 ff000000
000000ff 00000001 7faefaf0 00000000 00000005 7ff87c40 401548e8 00000000
40a52000 7faefaf0 00000005 7ff87c58 40155cc1 40a52030 00000005 00000000
Call Trace:
[<401058ca>] show_trace_log_lvl+0x1a/0x30
[<4010598b>] show_stack_log_lvl+0xab/0xd0
[<40105b7c>] show_registers+0x1cc/0x2d0
[<40105d96>] die+0x116/0x240
[<4011d7bb>] do_page_fault+0x18b/0x670
[<40721d22>] error_code+0x72/0x80
[<401548e8>] handle_IRQ_event+0x28/0x60
[<40155cc1>] handle_fasteoi_irq+0x71/0xe0
[<40107238>] do_IRQ+0x48/0xa0
[<401051fe>] common_interrupt+0x2e/0x40
[<401a344e>] proc_create+0x3e/0x120
[<401a3733>] proc_mkdir_mode+0x23/0x50
[<401a376f>] proc_mkdir+0xf/0x20
[<40156864>] register_handler_proc+0xd4/0xf0
[<401550c2>] setup_irq+0xe2/0x190
[<40155224>] request_irq+0xb4/0xd0
[<4058f524>] snd_gus_create+0x124/0x3c0
[<40aa4087>] snd_gusclassic_probe+0x2a7/0x4b0
[<403f5eff>] isa_bus_probe+0x1f/0x30
[<403f1944>] driver_probe_device+0x84/0x190
[<403f1a58>] __device_attach+0x8/0x10
[<403f0e63>] bus_for_each_drv+0x53/0x80
[<403f1b1b>] device_attach+0x8b/0x90
[<403f0dd8>] bus_attach_device+0x48/0x80
[<403efdbd>] device_add+0x45d/0x5a0
[<403eff12>] device_register+0x12/0x20
[<403f60c3>] isa_register_driver+0xb3/0x140
[<40aa3dd2>] alsa_card_gusclassic_init+0x12/0x20
[<40a665c3>] kernel_init+0x133/0x310
[<401054a7>] kernel_thread_helper+0x7/0x10
=======================
Code: Bad EIP value.
EIP: [<00000000>] 0x0 SS:ESP 0068:7ff87bf4
Kernel panic - not syncing: Fatal exception in interrupt

with these things fixed, i get the expected "no such hardware" result
from the driver initialization:

Calling initcall 0x40aa3dc0: alsa_card_gusclassic_init+0x0/0x20()
ALSA sound/isa/gus/gusclassic.c:136: [0x220] check 1 failed - 0xff
initcall 0x40aa3dc0: alsa_card_gusclassic_init+0x0/0x20() returned 0.
initcall 0x40aa3dc0 ran for 133 msecs:
alsa_card_gusclassic_init+0x0/0x20()

Signed-off-by: Ingo Molnar <mingo@xxxxxxx>
---
sound/isa/gus/gus_irq.c | 18 ++++++++++++------
sound/isa/gus/gus_main.c | 16 ++++++++--------
2 files changed, 20 insertions(+), 14 deletions(-)

Index: linux/sound/isa/gus/gus_irq.c
===================================================================
--- linux.orig/sound/isa/gus/gus_irq.c
+++ linux/sound/isa/gus/gus_irq.c
@@ -45,11 +45,13 @@ __again:
// snd_printk("IRQ: status = 0x%x\n", status);
if (status & 0x02) {
STAT_ADD(gus->gf1.interrupt_stat_midi_in);
- gus->gf1.interrupt_handler_midi_in(gus);
+ if (gus->gf1.interrupt_handler_midi_in)
+ gus->gf1.interrupt_handler_midi_in(gus);
}
if (status & 0x01) {
STAT_ADD(gus->gf1.interrupt_stat_midi_out);
- gus->gf1.interrupt_handler_midi_out(gus);
+ if (gus->gf1.interrupt_handler_midi_out)
+ gus->gf1.interrupt_handler_midi_out(gus);
}
if (status & (0x20 | 0x40)) {
unsigned int already, _current_;
@@ -85,20 +87,24 @@ __again:
}
if (status & 0x04) {
STAT_ADD(gus->gf1.interrupt_stat_timer1);
- gus->gf1.interrupt_handler_timer1(gus);
+ if (gus->gf1.interrupt_handler_timer1)
+ gus->gf1.interrupt_handler_timer1(gus);
}
if (status & 0x08) {
STAT_ADD(gus->gf1.interrupt_stat_timer2);
- gus->gf1.interrupt_handler_timer2(gus);
+ if (gus->gf1.interrupt_handler_timer2)
+ gus->gf1.interrupt_handler_timer2(gus);
}
if (status & 0x80) {
if (snd_gf1_i_look8(gus, SNDRV_GF1_GB_DRAM_DMA_CONTROL) & 0x40) {
STAT_ADD(gus->gf1.interrupt_stat_dma_write);
- gus->gf1.interrupt_handler_dma_write(gus);
+ if (gus->gf1.interrupt_handler_dma_write)
+ gus->gf1.interrupt_handler_dma_write(gus);
}
if (snd_gf1_i_look8(gus, SNDRV_GF1_GB_REC_DMA_CONTROL) & 0x40) {
STAT_ADD(gus->gf1.interrupt_stat_dma_read);
- gus->gf1.interrupt_handler_dma_read(gus);
+ if (gus->gf1.interrupt_handler_dma_read)
+ gus->gf1.interrupt_handler_dma_read(gus);
}
}
if (--loop > 0)
Index: linux/sound/isa/gus/gus_main.c
===================================================================
--- linux.orig/sound/isa/gus/gus_main.c
+++ linux/sound/isa/gus/gus_main.c
@@ -154,6 +154,14 @@ int snd_gus_create(struct snd_card *card
gus = kzalloc(sizeof(*gus), GFP_KERNEL);
if (gus == NULL)
return -ENOMEM;
+ spin_lock_init(&gus->reg_lock);
+ spin_lock_init(&gus->voice_alloc);
+ spin_lock_init(&gus->active_voice_lock);
+ spin_lock_init(&gus->event_lock);
+ spin_lock_init(&gus->dma_lock);
+ spin_lock_init(&gus->pcm_volume_level_lock);
+ spin_lock_init(&gus->uart_cmd_lock);
+ mutex_init(&gus->dma_mutex);
gus->gf1.irq = -1;
gus->gf1.dma1 = -1;
gus->gf1.dma2 = -1;
@@ -218,14 +226,6 @@ int snd_gus_create(struct snd_card *card
gus->gf1.pcm_channels = pcm_channels;
gus->gf1.volume_ramp = 25;
gus->gf1.smooth_pan = 1;
- spin_lock_init(&gus->reg_lock);
- spin_lock_init(&gus->voice_alloc);
- spin_lock_init(&gus->active_voice_lock);
- spin_lock_init(&gus->event_lock);
- spin_lock_init(&gus->dma_lock);
- spin_lock_init(&gus->pcm_volume_level_lock);
- spin_lock_init(&gus->uart_cmd_lock);
- mutex_init(&gus->dma_mutex);
if ((err = snd_device_new(card, SNDRV_DEV_LOWLEVEL, gus, &ops)) < 0) {
snd_gus_free(gus);
return err;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Anton Arapov: "Re: [PATCH] ipv4: kernel panic when only one unsecured port available"
Previous message: Stefan Richter: "Re: What still uses the block layer?"
Next in thread: Takashi Iwai: "Re: [patch] sound: fix bootup crash in snd_gus_interrupt()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]