sata_nv silently ignores discard (TRIM) command?

sata_nv silently ignores discard (TRIM) command?

[Alberto Mattea]

Hello all.
I'm testing linux 2.6.37 final with a Crucial RealSSD C300 (256GB).
The chipset is a nforce 570 (M2N32-SLI Deluxe); lspci:
IDE interface [0101]: nVidia Corporation MCP55 SATA Controller [10de:037f] 
(rev a2) (prog-if 85 [Master SecO PriO])
        Subsystem: ASUSTeK Computer Inc. Device [1043:cb84]
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- 
Stepping- SERR- FastB2B- DisINTx-
        Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- 
<TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0 (750ns min, 250ns max)
        Interrupt: pin C routed to IRQ 21
        Region 0: I/O ports at f200 [size=8]
        Region 1: I/O ports at f100 [size=4]
        Region 2: I/O ports at f000 [size=8]
        Region 3: I/O ports at ef00 [size=4]
        Region 4: I/O ports at ee00 [size=16]
        Region 5: Memory at fe02b000 (32-bit, non-prefetchable) [size=4K]
        Capabilities: [44] Power Management version 2
                Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA 
PME(D0-,D1-,D2-,D3hot-,D3cold-)
                Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
        Capabilities: [b0] MSI: Enable- Count=1/4 Maskable- 64bit+
                Address: 0000000000000000  Data: 0000
        Capabilities: [cc] HyperTransport: MSI Mapping Enable+ Fixed+
        Kernel driver in use: sata_nv
        Kernel modules: sata_nv

On the device there is a single ext4 partition, mounted with -o discard. 
Dmesg:
EXT4-fs (sda1): re-mounted. Opts: errors=remount-
ro,discard,data=writeback,commit=0

To test whether TRIM is working, I did the following:
# dd if=/dev/urandom of=tempfile count=100 bs=512k oflag=direct
# hdparm --fibmap tempfile
(File begins at 66090496)
# hdparm --read-sector 66090496 /dev/sda
(Random data)
# rm tempfile
# sync
# hdparm --read-sector 66090496 /dev/sda
Now, if I'm correct after a successful TRIM the last command should output all 
zeroes, but exactly the same random data is reported instead. However, there 
is nothing in dmesg about TRIM failing. Does sata_nv support TRIM on the 570?
Thanks
Alberto

Re: sata_nv silently ignores discard (TRIM) command?

[Jeff Moyer]

Alberto Mattea <alberto@mattea.info> writes:

> Hello all.
> I'm testing linux 2.6.37 final with a Crucial RealSSD C300 (256GB).
> The chipset is a nforce 570 (M2N32-SLI Deluxe); lspci:
> IDE interface [0101]: nVidia Corporation MCP55 SATA Controller [10de:037f] 
> (rev a2) (prog-if 85 [Master SecO PriO])
>         Subsystem: ASUSTeK Computer Inc. Device [1043:cb84]
>         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- 
> Stepping- SERR- FastB2B- DisINTx-
>         Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- 
> <TAbort- <MAbort- >SERR- <PERR- INTx-
>         Latency: 0 (750ns min, 250ns max)
>         Interrupt: pin C routed to IRQ 21
>         Region 0: I/O ports at f200 [size=8]
>         Region 1: I/O ports at f100 [size=4]
>         Region 2: I/O ports at f000 [size=8]
>         Region 3: I/O ports at ef00 [size=4]
>         Region 4: I/O ports at ee00 [size=16]
>         Region 5: Memory at fe02b000 (32-bit, non-prefetchable) [size=4K]
>         Capabilities: [44] Power Management version 2
>                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA 
> PME(D0-,D1-,D2-,D3hot-,D3cold-)
>                 Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
>         Capabilities: [b0] MSI: Enable- Count=1/4 Maskable- 64bit+
>                 Address: 0000000000000000  Data: 0000
>         Capabilities: [cc] HyperTransport: MSI Mapping Enable+ Fixed+
>         Kernel driver in use: sata_nv
>         Kernel modules: sata_nv
>
> On the device there is a single ext4 partition, mounted with -o discard. 
> Dmesg:
> EXT4-fs (sda1): re-mounted. Opts: errors=remount-
> ro,discard,data=writeback,commit=0
>
> To test whether TRIM is working, I did the following:
> # dd if=/dev/urandom of=tempfile count=100 bs=512k oflag=direct
> # hdparm --fibmap tempfile
> (File begins at 66090496)
> # hdparm --read-sector 66090496 /dev/sda
> (Random data)
> # rm tempfile
> # sync
> # hdparm --read-sector 66090496 /dev/sda
> Now, if I'm correct after a successful TRIM the last command should output all 
> zeroes, but exactly the same random data is reported instead. However, there 
> is nothing in dmesg about TRIM failing. Does sata_nv support TRIM on the 570?

What does 'cat /sys/block/sda/queue/discard_zeroes_data' say?

Cheers,
Jeff

Re: sata_nv silently ignores discard (TRIM) command?

[Alberto Mattea]

In data venerdì 14 gennaio 2011 19:32:45, Jeff Moyer ha scritto:
> 
> What does 'cat /sys/block/sda/queue/discard_zeroes_data' say?
> 
> Cheers,
> Jeff
Hi Jeff, it says 0. So I assume it's not working?
Thanks
Alberto

Re: sata_nv silently ignores discard (TRIM) command?

[Jeff Moyer]

Alberto Mattea <alberto@mattea.info> writes:

> In data venerdì 14 gennaio 2011 19:32:45, Jeff Moyer ha scritto:
>> 
>> What does 'cat /sys/block/sda/queue/discard_zeroes_data' say?
>> 
>> Cheers,
>> Jeff
> Hi Jeff, it says 0. So I assume it's not working?

That means that discarded blocks are not zeroed.  The trim was likely
processed just fine.

Cheers,
Jeff

Re: sata_nv silently ignores discard (TRIM) command?

[Alberto Mattea]

> 
> That means that discarded blocks are not zeroed.  The trim was likely
> processed just fine.
> 
> Cheers,
> Jeff

Thanks a lot! I was already worrying about changing motherboard.
Bye
Alberto