kexec kernel failed to restart aacraid device

Hi,

When try to kdump via sysrq-trigger, found kernel hung when try to restart aacraid device, kernel print below messages:

Loading aacraid.ko
AAC0: adapter kernel faile to start, init status = 0.

Normally startup, got following message what related the devices:

Adaptec aacraid driver 1.1-5[2461]
AAC0: kernel 4.2-0[8208] Jul 6 2006
AAC0: monitor 4.2-0[8208]
AAC0: bios 4.2-0[8208]
AAC0: serial C36FA5
AAC0: Non-DASD support enabled.

Device info from lspci:

02:03.0 0104: 9005:0285 (rev 01)
Subsystem: 9005:0286
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR+ FastB2B-
Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=slow >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 64 (250ns min, 250ns max), Cache Line Size: 64 bytes
Interrupt: pin A routed to IRQ 177
Region 0: Memory at f8000000 (32-bit, prefetchable) [size=64M]
Expansion ROM at fc9f0000 [disabled] [size=32K]
Capabilities: [80] Power Management version 2
Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Readed driver's codes found if ignore restart device at _aac_rx_init():

Changed

	if ((((status & 0x0c) != 0x0c) || aac_reset_devices || reset_devices) && 
		!aac_rx_restart_adapter(dev, 0))

to

	if ((aac_reset_devices || reset_devices) && 
		!aac_rx_restart_adapter(dev, 0))

and get rid of reset_devices from kdump command line, kexec kernel could bootup kernel normally, kdump would dump vmcore, looked like it caused by aac_rx_restart_adapter() could not restart devices, it made kdump kernel could not go on.

Any idea will appreciate~

Thanks,
Joe

Hi,

Checked aacraid drivers, initialzing and reset device at _aac_rx_init(), at the begin of function, set @adapter_enable_int to aac_rx_disable_interrupt(), when device status it KERNEL_PANIC, driver will try to reset device, before reset device, it will check device health or no by call aac_rx_check_health(), then aac_rx_check_health()->rx_sync_cmd() will use the aac_rx_disable_interrupt() to restore interrupt mask, is that the right action? if need to replace it to aac_rx_enable_interrupt_producer() before call aac_rx_check_health()?

Have attached the patch, please review it.

Thanks,
Joe

index f70d9f8..07142a6 100644
@@ -556,7 +556,12 @@ int _aac_rx_init(struct aac_dev *dev)
 	 */
 	status = rx_readl(dev, MUnit.OMRx[0]);
 	if (status & KERNEL_PANIC) {
-		if (aac_rx_restart_adapter(dev, aac_rx_check_health(dev)))
+		int health;
+
+		dev->a_ops.adapter_enable_int = aac_rx_enable_interrupt_producer;
+		health = aac_rx_check_health(dev);
+		dev->a_ops.adapter_enable_int = aac_rx_disable_interrupt;
+		if (aac_rx_restart_adapter(dev, health))
 			goto error_iounmap;
 		++restart;
 	}

@@ -594,7 +599,12 @@ int _aac_rx_init(struct aac_dev *dev)
 		  ((startup_timeout > 60)
 		    ? (startup_timeout - 60)
 		    : (startup_timeout / 2))))) {
-			if (likely(!aac_rx_restart_adapter(dev, aac_rx_check_health(dev))))
+			int health;
+
+			dev->a_ops.adapter_enable_int = aac_rx_enable_interrupt_producer;
+			health = aac_rx_check_health(dev);
+			dev->a_ops.adapter_enable_int = aac_rx_disable_interrupt;
+			if (likely(!aac_rx_restart_adapter(dev, health)))
 				start = jiffies;
 			++restart;
 		}