2019-05-01 21:47:46

by Stephen Rothwell

[permalink] [raw]
Subject: linux-next: Signed-off-by missing for commit in the parisc-hd tree

Hi all,

Commit

2b731a920986 ("parisc: Update huge TLB page support to use per-pagetable spinlock")

is missing a Signed-off-by from its committer.

--
Cheers,
Stephen Rothwell


Attachments:
(No filename) (499.00 B)
OpenPGP digital signature

2019-05-02 05:50:02

by Helge Deller

[permalink] [raw]
Subject: Re: linux-next: Signed-off-by missing for commit in the parisc-hd tree

On 01.05.19 23:45, Stephen Rothwell wrote:
> Commit
> 2b731a920986 ("parisc: Update huge TLB page support to use per-pagetable spinlock")
> is missing a Signed-off-by from its committer.

Thanks. Fixed now.
Helge


Attachments:
signature.asc (235.00 B)
OpenPGP digital signature

2019-05-12 20:51:15

by cp

[permalink] [raw]
Subject: C3600, sata controller

guys,
I asked a friend to lend us his card, which is the card2 listed below
and not things are becoming really weird

card1, Silicon Image, Inc. Adaptec AAR-1210SA
card2, Silicon Image, Inc. SiI 3124 PCI-X Serial ATA

the card1 is a PCI32 bit card, and uses the "sata_sil" driver
the card2 is a PCI64 bit card, and uses the "sata_sil24" driver

card1 is installed in a PCI32 slot, and it's moving 32Gbyte in a loop
without any issue

I have just tested card2 in each of the PCI-X slot (including the 3.3V
one) and ... it has always triggered the HPMC unit, crashing the
machine

the hardware is different, and drivers are also different, but ....
maybe the problem is related to 32 vs 64bit?

I am going to repeat the test with card2 installed in a PCI32 slot.
This should force it to 32bit: will it work correctly?
(hope this makes sense)

---

we have also checked if the kernel is correctly handling the ram
we have a C3600 with 8Gbyte of ram

mount -t tmpfs -o size=7G tmpfs /mnt/ramdrive/
dd if=/dev/zero of=/mnt/ramdrive/test.bin
badblocks -swv /mnt/ramdrive/test.bin

checking for bad blocks in read-write mode
From block 0 to 7340031
Testing with pattern 0xaa: done
Reading and comparing: done
Testing with pattern 0x55: done
Reading and comparing: done
Testing with pattern 0xff: done
Reading and comparing: done
Testing with pattern 0x00: done
Reading and comparing: done
Pass completed, 0 bad blocks found.

no problems found

2019-05-12 20:52:50

by cp

[permalink] [raw]
Subject: Re: C3600, sata controller

kernel's config and dmesg are here

http://www.downthebunker.com/chunk_of/stuff/public/projects/workstations/hppa/c3600/kernel/5.1-git-v5.2-64bit/

http://www.downthebunker.com/chunk_of/stuff/public/projects/workstations/hppa/c3600/kernel/4.16.2-64bit/

Il giorno dom 12 mag 2019 alle ore 22:44 Carlo Pisani
<[email protected]> ha scritto:
>
> guys,
> I asked a friend to lend us his card, which is the card2 listed below
> and not things are becoming really weird
>
> card1, Silicon Image, Inc. Adaptec AAR-1210SA
> card2, Silicon Image, Inc. SiI 3124 PCI-X Serial ATA
>
> the card1 is a PCI32 bit card, and uses the "sata_sil" driver
> the card2 is a PCI64 bit card, and uses the "sata_sil24" driver
>
> card1 is installed in a PCI32 slot, and it's moving 32Gbyte in a loop
> without any issue
>
> I have just tested card2 in each of the PCI-X slot (including the 3.3V
> one) and ... it has always triggered the HPMC unit, crashing the
> machine
>
> the hardware is different, and drivers are also different, but ....
> maybe the problem is related to 32 vs 64bit?
>
> I am going to repeat the test with card2 installed in a PCI32 slot.
> This should force it to 32bit: will it work correctly?
> (hope this makes sense)
>
> ---
>
> we have also checked if the kernel is correctly handling the ram
> we have a C3600 with 8Gbyte of ram
>
> mount -t tmpfs -o size=7G tmpfs /mnt/ramdrive/
> dd if=/dev/zero of=/mnt/ramdrive/test.bin
> badblocks -swv /mnt/ramdrive/test.bin
>
> checking for bad blocks in read-write mode
> From block 0 to 7340031
> Testing with pattern 0xaa: done
> Reading and comparing: done
> Testing with pattern 0x55: done
> Reading and comparing: done
> Testing with pattern 0xff: done
> Reading and comparing: done
> Testing with pattern 0x00: done
> Reading and comparing: done
> Pass completed, 0 bad blocks found.
>
> no problems found

2019-05-12 21:17:30

by cp

[permalink] [raw]
Subject: Re: C3600, sata controller

> The c3600 doesn't have any PCI-X slots (only PCI) as far as I can tell from user manuals.

PCI-32/33 device I/O bus
PCI-64/33 high-performance device I/O bus <----------------- this is
PCI-X, 64bit 5V
PCI-64/66 high-performance graphics I/O bus <----------------- this is
PCI-X, 64bit 3.3V

https://www.openpa.net/systems/hp-visualize_b1000_c3000_c3600.html

2019-05-12 21:19:51

by John David Anglin

[permalink] [raw]
Subject: Re: C3600, sata controller

The c3600 doesn't have any PCI-X slots (only PCI) as far as I can tell from user manuals.

The SY-PCX40009 PCI-X card that I have says it supports two models: 32-bit at 66 MHz and 64-bit
at 133 MHz.  The 32-bit slots in c3600 are 33 MHz.  Whether the card will work at 33 MHz is not
clear.  The card says it's backward compatible to PCI 2.3.

On 2019-05-12 4:44 p.m., Carlo Pisani wrote:
> guys,
> I asked a friend to lend us his card, which is the card2 listed below
> and not things are becoming really weird
>
> card1, Silicon Image, Inc. Adaptec AAR-1210SA
> card2, Silicon Image, Inc. SiI 3124 PCI-X Serial ATA
>
> the card1 is a PCI32 bit card, and uses the "sata_sil" driver
> the card2 is a PCI64 bit card, and uses the "sata_sil24" driver
>
> card1 is installed in a PCI32 slot, and it's moving 32Gbyte in a loop
> without any issue
>
> I have just tested card2 in each of the PCI-X slot (including the 3.3V
> one) and ... it has always triggered the HPMC unit, crashing the
> machine
>
> the hardware is different, and drivers are also different, but ....
> maybe the problem is related to 32 vs 64bit?
>
> I am going to repeat the test with card2 installed in a PCI32 slot.
> This should force it to 32bit: will it work correctly?
> (hope this makes sense)
>
> ---
>
> we have also checked if the kernel is correctly handling the ram
> we have a C3600 with 8Gbyte of ram
>
> mount -t tmpfs -o size=7G tmpfs /mnt/ramdrive/
> dd if=/dev/zero of=/mnt/ramdrive/test.bin
> badblocks -swv /mnt/ramdrive/test.bin
>
> checking for bad blocks in read-write mode
> >From block 0 to 7340031
> Testing with pattern 0xaa: done
> Reading and comparing: done
> Testing with pattern 0x55: done
> Reading and comparing: done
> Testing with pattern 0xff: done
> Reading and comparing: done
> Testing with pattern 0x00: done
> Reading and comparing: done
> Pass completed, 0 bad blocks found.
>
> no problems found
>


--
John David Anglin [email protected]

2019-05-13 16:39:16

by cp

[permalink] [raw]
Subject: Re: C3600, sata controller

hi, good news!
after a burn-in test of 8hours, we can affirm that Adaptec 1210SA does work
The chip is "Silicon Image Sil31122", so the kernel module is
"sata_sil" (not sata_sil24)

while [ 1 ]
do
for item in `ls *.bin`
do
rm -f $copy.out
echo -n "$item ... "
mycp $item $copy.out
echo "done"
done
done

*.bin are giant files, 1GB, 2GB, 4GB, 8GB, 16GB, 32GB

it has been tested on a PCI 32bit lost of a C3600
with kernel /v4.16!!!

"mycp" does a copy and then checks the md5 checksum

I think this card can be added to the list!

Our list has been already updated
http://www.downthebunker.com/reloaded/space/viewtopic.php?f=50&p=1632

2019-05-13 23:15:13

by John David Anglin

[permalink] [raw]
Subject: Re: C3600, sata controller

On 2019-05-12 5:15 p.m., Carlo Pisani wrote:
>> The c3600 doesn't have any PCI-X slots (only PCI) as far as I can tell from user manuals.
> PCI-32/33 device I/O bus
> PCI-64/33 high-performance device I/O bus <----------------- this is
> PCI-X, 64bit 5V
> PCI-64/66 high-performance graphics I/O bus <----------------- this is
> PCI-X, 64bit 3.3V
>
> https://www.openpa.net/systems/hp-visualize_b1000_c3000_c3600.html
I think that you are wrong.  HP would have said the slots were PCI-X compatible if they were.
They did in c8000.

The issue is probably the signalling voltage and clock rate capability.  33 MHz always uses 5V
signalling whereas PCX-X cards typically support 66 and 133 MHz.  Thus, I think PCI-X cards use
CMOS 3.3V signalling although they may accept 5V on their inputs.  If the c3600 uses older
TTL logic, it is likely that 3.3V CMOS logic can't provide the high level for 5V TTL logic.

I suspect the only slot that might work with a PCI-X card is the 3.3V SL2.  However, if you see I/O errors
in the PIM dump following a HPMC, then it's not compatible.

So, probably you would have better luck with the PCI Syba SD-SATA150R in c3600.  However, it is similar
to the Adaptec 1210SA which works.  Maybe they differ in RAID capability.

Dave

--
John David Anglin [email protected]

2019-05-25 09:00:50

by cp

[permalink] [raw]
Subject: Re: C3600, sata controller

guys?
yahoooooooooooooooo, SYBA-SY-PCX40009 does woooork :D :D

02:03.0 RAID bus controller: Silicon Image, Inc. SiI 3124 PCI-X Serial
ATA Controller (rev 01)

10 hours burn-in test passed!!!!

while [ 1 ]
do
for item in `ls *.bin`
do
rm -f $copy.out
echo -n "$item ... "
time cp_and_check_md5sum $item $copy.out
echo "done"
done
done

kernel 5.1, from the git repository, compiled with SMP

dmesg | grep altern
[ 2.551002] alternatives: applied 156 out of 175 patches

only applied this patch:

--- a/arch/parisc/include/asm/cache.h
+++ b/arch/parisc/include/asm/cache.h
@@ -52,7 +52,6 @@ void parisc_setup_cache_timing(void);

#define asm_io_fdc(addr) asm volatile("fdc %%r0(%0)" \
ALTERNATIVE(ALT_COND_NO_DCACHE, INSN_NOP) \
- ALTERNATIVE(ALT_COND_NO_IOC_FDC, INSN_NOP) \
: : "r" (addr) : "memory")
#define asm_io_sync() asm volatile("sync" \
ALTERNATIVE(ALT_COND_NO_DCACHE, INSN_NOP) \