Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Alan Cox]

On Llu, 2005-11-07 at 21:04 +0100, Johan Palmqvist wrote:
> When used with a HighPoint RocketRAID 1520 SATA (HPT372N) the hpt366 
> driver, compiled as a module, oops'es on loading. If the driver is 
> compiled into the kernel it causes a kernel panic on boot while 
> detecting the card. Kernels tested: 2.6.13.2, 2.6.13.4 and 2.6.14. 
> Please CC any answers to me since I'm not on the list.

I'm working on this with the new libata drivers. The list of what is
broken in the old driver for 372N/302N is rather large as they are
actually very different to the other chips.

I've got the correct timing tables and PLL tune information for the 302N
now which appears to be very similar if not identical for these needs so
if someone wants to hack the old drivers/ide/pci driver a bit I can
provide reasonably accurate chip id and chip pll base information to
enable a fix.

PS: Anyone got a 371N (specifically 371N) can send me the pci id and
rev.

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Sergei Shtylylov]

Hello.

Alan Cox wrote:

> I've got the correct timing tables and PLL tune information for the 302N
> now which appears to be very similar if not identical for these needs so
> if someone wants to hack the old drivers/ide/pci driver a bit I can
> provide reasonably accurate chip id and chip pll base information to
> enable a fix.

    I'm currently working on the old driver to bring in the HPT371N support 
and fix HPT371 clock setup (and make the other chip's clock setups correct 
along the way). Unfortunately I can test only on 371N and 370A, the board with 
371 soldered in has PCI contreoller fried.

WBR, Sergei

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Sergei Shtylyov]

Hello.

Sergei Shtylylov wrote:

> Alan Cox wrote:

>> I've got the correct timing tables and PLL tune information for the 302N
>> now which appears to be very similar if not identical for these needs so
>> if someone wants to hack the old drivers/ide/pci driver a bit I can
>> provide reasonably accurate chip id and chip pll base information to
>> enable a fix.

>    I'm currently working on the old driver to bring in the HPT371N 
> support and fix HPT371 clock setup (and make the other chip's clock 
> setups correct along the way). Unfortunately I can test only on 371N and 
> 370A, the board with 371 soldered in has PCI contreoller fried.

    The driver rewrite is almost complete (after being preempted several 
times), I'm awaiting it to also be tested on HPT302. The changes are HUGE, the 
patch is about 70 KB in size...
    Meanwhile, I've noticed (just now :-) a bug in HPT372N clock turnaround 
code (it accesses extra DMA ports relative to dma_base, not dma_master which 
would make it do BAD things if called on a secondary channel) -- guess I 
should post them in a separate patch?

WBR, Sergei

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Sergei Shtylyov]

Hello.

Alan Cox wrote:

>>When used with a HighPoint RocketRAID 1520 SATA (HPT372N) the hpt366 
>>driver, compiled as a module, oops'es on loading. If the driver is 
>>compiled into the kernel it causes a kernel panic on boot while 
>>detecting the card. Kernels tested: 2.6.13.2, 2.6.13.4 and 2.6.14. 
>>Please CC any answers to me since I'm not on the list.

    I think I've dealt with this oops now, see my recent patches to this
driver, particularly the one that reads the f_CNT value saved by BIOS...

WBR, Sergei

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Sergei Shtylyov]

Hello.

Alan Cox wrote:
> On Iau, 2006-05-04 at 23:53 +0400, Sergei Shtylyov wrote:
> 
>>    I think I've dealt with this oops now, see my recent patches to this
>>driver, particularly the one that reads the f_CNT value saved by BIOS...
> 
> 
> I have them queued to review in depth, although in that specific case I
> don't know if I can help - none of my cards set the BIOS provided
> value..

    Judging on a FREQ value (about 125) from the boot logs, the BIOS had set 
up PLL to 50 MHz (125/192 ~ 32/50) which should have "spoilt" the value in the 
f_CNT register. Hence, we may guess that the BIOS had saved the f_CNT value 
beforehand. If not, there's not much we can do except to guess at the DPLL 
frequence that BIOS might have set.

MBR, Sergei

Re: hpt366 driver oops or panic with HighPoint RocketRAID 1520 SATA (HPT372N)

[Alan Cox]

On Iau, 2006-05-04 at 23:53 +0400, Sergei Shtylyov wrote:
>     I think I've dealt with this oops now, see my recent patches to this
> driver, particularly the one that reads the f_CNT value saved by BIOS...

I have them queued to review in depth, although in that specific case I
don't know if I can help - none of my cards set the BIOS provided
value..