RFC: ARM related fixes - GIC, cache line size, PCI FDT & AHCI

[Warner Losh]

Just one comment below:

On Dec 19, 2013, at 4:23 AM, Zbigniew Bodek wrote:

> Hello Warner,
> 
> Thanks for your review.
> Please see below.
> 
> Best regards
> zbb
> 
> On 16.12.2013 22:50, Warner Losh wrote:
>> 
>> On Dec 16, 2013, at 10:48 AM, Zbigniew Bodek wrote:
>> 
>>> Hello Everyone.
>>> 
>>> We would like to submit some new patches recently developed by Semihalf.
>> 
>> Thanks for your efforts here. Would there be a need to MFC these patches to 10 and/or 9?
> [ZBB] We can take a look at that after committing to current.
>> 
>>> You can find them here:
>>> http://people.freebsd.org/~zbb/Semihalf/12.2013/
>>> 
>>> Detailed description is available in the commit logs but in general:
>>> 
>>> -- 0001-Resolve-cache-line-size-using-CP15.patch
>>>  - use cache line size acquired in runtime
>> 
>> This looks good (assuming the register definitions are good, I didn't verify them). Might be useful to publish these values to userland somehow, maybe via a sysctl. This is option, but would be nice to have.
>> 
>> Would also be good to have USB_ALIGN kassert if using USB and this value is < that macro.
>> 
>> Do other, older CPUs need similar decoding, or are the hard-wired values good?
> [ZBB] I have no knowledge about that. Someone who is using older ARMs
> should take a look.

I'll try to take a look, but I only have the armv5 gear to look at. I'm most concerned about RPi...

Warner

>> 
>>> -- 0002-GIC-polarity-and-level-support.patch
>>>  - suport for setting trigger level and polarity in GIC
>> 
>> This one seems to be OK, but is complicated enough that more eyeballs and testing would be good....
> [ZBB] OK. Someone is testing this right now.
>> 
>>> -- 0003-Add-PCI-FDT-interrupt-trigger-polarity-parsing.patch
>>>  - trigger and polarity parsing for PCI FDT interrupts
>> 
>> This patch might conflict.
>> 
>> +		if (trig)
>> +			*trig = tmp_trig;
>> +		if (pol)
>> +			*pol = tmp_pol;
>> 
>> Are not zero values also valid?
> [ZBB] Yeah. We will skip this patch for now.
>> 
>>> -- 0004-Do-not-attach-to-bridges-in-AHCI-driver.patch
>> 
>> I'd add the word 'PCI' before bridge in the comment, otherwise this looks good.
> [ZBB] OK.
>> 
>>> -- 0005-Use-only-mapped-BIOs-on-ARM.patch
>> 
>> A comment about what MVA is would be useful.. Otherwise does what it says.
> [ZBB]. This should refer to just Virtual Address (not "Modified") so we
> will change it back to VA.
>> 
>>> - Two patches enabling the AHCI driver on ARM chips
>> 
>> I didn't see these. Did I miss something? Or is the unmapped stuff all it takes?
> [ZBB] Yes, I meant the two above(0004 && 0005).
>> 
>>> We will appreciate if you could post your comments and/or remarks by the
>>> end of this week when we plan to commit the changes.
>>> 
>>> Best regards
>>> zbb
>>> _______________________________________________
>>> freebsd-arm at freebsd.org mailing list
>>> http://lists.freebsd.org/mailman/listinfo/freebsd-arm
>>> To unsubscribe, send any mail to "freebsd-arm-unsubscribe at freebsd.org"
>> 
>