Gentoo Archives: gentoo-user

From: Michael Mol <mikemol@×××××.com>
To: gentoo-user@l.g.o
Subject: Re: [gentoo-user] Re: Hard drive RPMs and data speed.
Date: Mon, 31 Oct 2011 16:54:15
In Reply to: Re: [gentoo-user] Re: Hard drive RPMs and data speed. by Alex Schuster
1 On Mon, Oct 31, 2011 at 12:21 PM, Alex Schuster <wonko@×××××××××.org> wrote:
2 > Michael Mol writes:
3 >
4 >> My point is that the numbers aren't what mattered here. My point is
5 >> that SAMSUNG sold me a shoddy product, replaced it with another
6 >> instance of the the same shoddy product, wouldn't replace it again,
7 >> and never addressed a detailed technical report of a systemic problem
8 >> in the same. Bad tech, bad customer service, and it looked like this
9 >> was a more common scenario than among other manufacturers. All of it
10 >> boiled down to a nasty case of being a bad candidate for spending time
11 >> and money.
12 >
13 > Samsung, uh? Here's my story of today. My fried just bought two external
14 > USB drives. I wanted to know which brand the HD is, so I checked with
15 > hdparm -I, and googled for SAMSUNG HD204UI. I found a story about a bug
16 > which makes the drive sometimes forget to write a block when it is
17 > attached to a SATA adapter in AHCI mode and when the ATA command
18 > "IDENTIFY DEVICE" is sent (like in hdparm -I or when using the
19 > smartmontools). There is a firmware patch for this, this is good. But on
20 > the annoying side:
22 That could very likely be the nature of the initial symptom for my
23 second failed drive. I recall being angry about silent corruption,
24 with SMART not reporting anything interesting. Drive failed
25 differently later on, IIRC. I still have it in the same eSATA external
26 enclosure I was using at the time. I'll have to look.
28 >
29 > - You need to make a DOS boot floppy and copy the patch there. I don't
30 >  know how exactly to do this, and I read about people using Linux who
31 >  needed over an hour for this or even failed. Can't they just let me
32 >  download an image I can boot from?
34 Any idea if it works from FreeDOS?
36 >
37 > - It doesn't work over USB, so I would have to install the drive in a PC.
39 Does the enclosure doens't contain an eSATA port? That's almost
40 certain to be a direct passthrough.
42 >
43 > - The new firmware has exactly the same revision number. How stupid is
44 >  this?? I cannot even find out whether the drives have the problem or
45 >  not. Except by trying to reproduce the problem.
47 Sounds like the only way you can be certain the drives don't have the
48 problem is by installing the patched firmware.
50 >
51 > Here's a link to the but I described, but It's German only.
52 >
53 > I also read some angry comments about Samsung there. Question is, are
54 > other manufacturers better? And wasn't Samsung Electronics bought by
55 > Seagate anyway?
56 >
57 >
58 > Any idea whether an external USB drive case might count as a SATA
59 > controller in AHCI mode? I tried to trigger the bug, but that did not
60 > happen, so I guess it's fine, at least when being in the USB case.
62 The drive inside is SATA, so the USB enclosure is translating USB
63 mass-storage commands to commands the SATA drive can understand.
64 AFAIK, AHCI vs Legacy mode is a function of the SATA *controller*, not
65 of the drive itself; legacy mode takes an older protocol, converts it
66 to SATA commands, and then dispatches those SATA commands. I'll
67 venture a guess that when going through Legacy mode, whatever commands
68 trigger the bug aren't used in the Legacy->SATA conversion, whereas
69 AHCI, with its closer-to-metal nature, exposes those commands for use.
70 Whether or not the USB enclosure will trigger those bugs would depend
71 on whether or not the USB Mass Storage<->SATA translation uses those
72 commands or not.
74 (In my mind, this is feels like the old AGP->PCI Express transition.
75 Early PCIe video cards were actually AGP cards with an AGP->PCIe
76 bridge/adapter chip onboard, because it was faster and cheaper to get
77 to market by throwing an adapter component in the sequence. However, I
78 don't know enough about the ASIC market for USB hard drive enclosures
79 to know whether chips with adapter layers (like that legacy->SATA
80 command translation, but at a hardware level, making it
81 USB->legacy->SATA) will be the cheaper part to source than chips which
82 convert more directly between the USB Mass Storage and SATA
83 protocols.)
86 >
87 > Another problem is that data access frequently stalls on her PC, like when
88 > transferring data or doing a mke2fs. After a while, this message appears
89 > in syslog, and the process continues for a while, until it happens again:
90 >
91 > usb 1-4: reset high speed USB device using ehci_hcd and address 7
93 That looks incredibly familiar. I kept getting those, and then
94 switched to my enclosure's eSATA port. That's when the drive started
95 giving me different problems. At the time, I'd assumed it was the
96 enclosure's USB components at fault.
98 >
99 > Same problem with a GRML boot cd and on another USB port. Happens with
100 > both drives. But it is fine on my PC.
103 It sounds like the drives might be salvageable with a firmware patch,
104 now. I'd suggest extracting the drives, plugging them into a PC,
105 updating the firmware, and then putting them back in the enclosure.
106 That way, you're certain the drives don't still have the known-buggy
107 firmware, at the expense of some labor.
109 --
110 :wq