On Tuesday, September 19, 2006 7:17 PM, James Bottomley wrote:
> Well, by standard, I'm supposed to start out negotiation with the
> fastest I can possibly do and the drive modifies this in its
> reply. The
> heuristics over DT vs non-DT in the transport class are to cope with
> arguably buggy drives in the field. I've never come across a
> DT capable
> drive on an LVD bus that showed a problem in this regard ... have you
> found one?
>
Yes I have a U160 drive that does that. Meaning sending a ppr
with factor=0x8, the device returns factor=0xA. Instead
if you send a factor=0x9, the device returns factor=0x9. In
other words, spi transport is short changing nego on U160 devices.
See my proof at the end of this email.
My suggestion is:
the spi transport layer should only negotiate at factor=0x8
when both QAS and IUS are enabled in byte 56 of the inquiry
data. When you see both these bits zero, you should start
your max speed inquiry test at factor=0x9 instead of factor=0x8.
Yeah, I know that this came out in SPI-3, but there really
are no U160 devices out there that can actually do QAS.
Also, another problem is you should only enable QAS when all
the devices on a single bus are U320(excluding safte proc).
Meaning, if you have mixed mode case, such as U160 mixed with
U320, you should disable QAS for all the U320 devices.
The problem you will face is the U320 devices with QAS enabled
are going to starve the bus, and the other devices never win
arbitration. I've seen in my test bed using current
spi transport. In my environment, I have two U160 devices,
and one U320 device. Spi transport will enable
QAS for the U320 device.
-----------------------------------------------------------------
Here is my proof. Here are two scsi bus traces.
(1) The first trace shows that sending ppr with
factor=0x8 nego, the device returns with factor=0xA.
(2) The second trace shows that sending ppr with
factor=0x9 nego, the device returns with factor=0x9.
Fist Trace:
129 2_414 Arb
130 1_316 Arbwin 7
131 0_788 +Select 7,0
132 11_878 +Sel/Resel End
133 178_276 +MsgOut C0 Identify
20 Simple Queue
01 Tag
01 06 04 08 00 7F 01 D3
Parallel Protocol Pw:6.25ns,off:127,Width:16
bits,
IU=1 DT=1 QAS=0
134 44_228 MsgIn 01 06 04 0A 00 3F 01 00
Parallel Protocol Pw:25ns,off:63,Width:16 bits,
IU=0 DT=0 QAS=0
135 197_962 CMD - Inquiry
12 00 00 00 60 00
136 1_174 DataIn (W)
0h: 00 00 03 02 5B 00 00 3E ....[..>
8h: 51 55 41 4E 54 55 4D 20 QUANTUM
10h: 41 54 4C 41 53 20 56 20 ATLAS V
18h: 33 36 20 57 4C 53 20 20 36 WLS
20h: 30 41 30 36 31 34 33 39 0A061439
28h: 39 31 31 32 32 30 32 38 91122028
30h: 20 20 39 39 31 30 30 37 991007
38h: 0C 30 30 30 30 30 30 30 .0000000
40h: 30 30 30 30 30 30 30 30 00000000
48h: 30 30 30 30 30 30 30 30 00000000
50h: 30 30 30 30 30 30 30 30 00000000
58h: 30 30 30 30 30 30 30 30 00000000
(96 Bytes)
137 46_262 DEnd 96 Bytes
138 0_938 Status 00 Good
139 1_224 MsgIn 00 Cmd Complete
140 59_330 Bus Free
Second Trace:
128 730_298 Bus Free
129 2_412 Arb
130 1_314 Arbwin 7
131 0_790 +Select 7,0
132 12_830 +Sel/Resel End
133 161_576 +MsgOut C0 Identify
20 Simple Queue
01 Tag
01 06 04 09 00 7F 01 02
Parallel Protocol Pw:12.5ns,off:127,Width:16
bits,
IU=0 DT=1 QAS=0
134 45_576 MsgIn 01 06 04 09 00 3F 01 02
Parallel Protocol Pw:12.5ns,off:63,Width:16 bits,
IU=0 DT=1 QAS=0
135 201_852 CMD - Inquiry
12 00 00 00 60 00
136 0_644 DataIn (DT)
0h: 00 00 03 02 5B 00 00 3E ....[..>
8h: 51 55 41 4E 54 55 4D 20 QUANTUM
10h: 41 54 4C 41 53 20 56 20 ATLAS V
18h: 33 36 20 57 4C 53 20 20 36 WLS
20h: 30 41 30 36 31 34 33 39 0A061439
28h: 39 31 31 32 32 30 32 38 91122028
30h: 20 20 39 39 31 30 30 37 991007
38h: 0C 30 30 30 30 30 30 30 .0000000
40h: 30 30 30 30 30 30 30 30 00000000
48h: 30 30 30 30 30 30 30 30 00000000
50h: 30 30 30 30 30 30 30 30 00000000
58h: 30 30 30 30 30 30 30 30 00000000
(96 Bytes)
137 30_054 DEnd 96 Bytes
138 0_938 Status 00 Good
139 1_204 MsgIn 00 Cmd Complete
140 88_910 Bus Free
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
