Never mind, I found the issue

hw.ata.ata_dma=0
hw.ata.atapi_dma=0

in /boot/loader.conf, hanging around from the days of wood burning power
supplies was causing this strange behaviour!  Sorry for the noise folks!

        ---Mike

On 8/22/2018 12:17 PM, Mike Tancsa wrote:
> OK,
>       Some more odd things going on. If I write to a file from dd some random
> junk, not all of the file is saved when I do an unmount. This seems to
> be specific either to the ata controller of the APU or to the msata disk
> as I tested on a regular server with plain old disks and no issue
> 
> Eric, any chance you can try this on your APU as well ?
> 
> 
> 0# umount /mnt
> 0# mount /dev/ada0p1 /mnt
> 0# dd if=/dev/urandom of=/mnt/junk2 bs=512k count=4
> 4+0 records in
> 4+0 records out
> 2097152 bytes transferred in 0.101394 secs (20683253 bytes/sec)
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = acca1d8997c3b9d906b40d99cba734b6
> 0# hd /mnt/junk2 > /tmp/hd-junk2
> 0# ls -l /mnt/junk2
> -rw-------  1 root  wheel  - 2097152 Aug 22 11:53 /mnt/junk2
> 0# umount /mnt
> 0# mount /dev/ada0p1 /mnt
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = 2ef0cadf32db4e07a8edf4fc66f8a4eb
> 0# ls -l /mnt/junk2
> -rw-------  1 root  wheel  - 2097152 Aug 22 11:53 /mnt/junk2
> 0# hd /mnt/junk2 > /tmp/hd-junk2-post-mount
> 0# ls -l /tmp/hd-junk2*
> -rw-------  1 root  wheel  - 10354697 Aug 22 11:54 /tmp/hd-junk2
> -rw-------  1 root  wheel  -  1941594 Aug 22 11:54 /tmp/hd-junk2-post-mount
> 0#
> 
> If I do the same test on a USB stick it works as expected
> 
> 0# umount /mnt
> 0# mount /dev/da0p1 /mnt
> 0# dd if=/dev/urandom of=/mnt/junk2 bs=512k count=4
> 4+0 records in
> 4+0 records out
> 2097152 bytes transferred in 0.102256 secs (20508777 bytes/sec)
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = 51190332cdd4ef898bdc9c2520a9f749
> 0# umount /mnt
> 0# mount /dev/da0p1 /mnt
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = 51190332cdd4ef898bdc9c2520a9f749
> 0#
> Same with an SD card. All is OK
> 
> 0# umount /mnt
> 0# mount /dev/mmcsd0s2a /mnt
> 0# dd if=/dev/urandom of=/mnt/junk2 bs=512k count=4
> 4+0 records in
> 4+0 records out
> 2097152 bytes transferred in 0.107400 secs (19526588 bytes/sec)
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = 0811af4e57ab5a3bc224e5b8f8b3bc29
> 0# umount /mnt
> 0# mount /dev/mmcsd0s2a /mnt
> 0# md5 /mnt/junk2
> MD5 (/mnt/junk2) = 0811af4e57ab5a3bc224e5b8f8b3bc29
> 0#
> 
> Looking at hd before and after, its missing the end of the file
> 
> --- hd-junk2    2018-08-22 11:54:09.891572000 -0400
> +++ hd-junk2-post-mount 2018-08-22 11:54:41.996983000 -0400
> @@ -24574,106500 +24574,6 @@
>  0005ffd0  8b 71 87 b8 78 03 72 ca  0e 06 3b d4 31 fd 18 f8
> |.q..x.r...;.1...|
>  0005ffe0  de 39 72 fb c8 39 fd 1f  93 75 10 de 05 56 43 fb
> |.9r..9...u...VC.|
>  0005fff0  40 ce 54 e2 a4 17 3e 1e  ec 01 b7 fd 1b 69 b7 6f
> |@.T...>......i.o|
> -00060000  91 f3 02 ea 95 f4 12 1c  bf 00 68 1b 3d 8c 01 43
> |..........h.=..C|
> -00060010  f6 5b 4e ec 7f 37 19 15  5b c4 e6 fb 88 27 1c 54
> |.[N..7..[....'.T|
> -00060020  15 6c 02 7d fb 00 06 c1  4a 4a bf ce 9a 1b fe d4
> |.l.}....JJ......|
> -00060030  1c 3c b5 05 b6 4f 4e 62  b5 03 e3 e7 5e 27 d6 71
> |.<...ONb....^'.q|
> -00060040  ea 22 00 99 9d 13 e8 a9  64 0e fd 13 cc 23 73 67
> |."......d....#sg|
> -00060050  8e 78 0a ad ae 70 ab e4  22 b4 b7 b9 3b 75 9f 85
> |.x...p.."...;u..|
> -00060060  53 39 0c af 15 39 5f 04  ac 3c 65 e9 ea 29 1d b7
> |S9...9_..<e..)..|
> -00060070  ea f0 2f 19 3c 6d 1c 21  f1 58 4a 4b a8 26 8e f6
> |../.<m.!.XJK.&..|
> -00060080  05 99 8a 9d 54 75 e4 77  78 78 6c 75 21 31 d4 0c
> |....Tu.wxxlu!1..|
> -00060090  52 88 c6 65 c0 09 04 ce  7f 5f 29 0c 46 9a 68 13
> |R..e....._).F.h.|
> -000600a0  73 30 97 78 d7 b7 d2 ba  8f 73 27 58 3d eb 0c d6
> |s0.x.....s'X=...|
> 
> 
> 1# camcontrol identify ada0
> pass0: <SATA SSD S9FM02.0> ACS-3 ATA SATA 3.x device
> pass0: 600.000MB/s transfers (SATA 3.x, PIO4, PIO 8192bytes)
> 
> protocol              ATA/ATAPI-10 SATA 3.x
> device model          SATA SSD
> firmware revision     S9FM02.0
> serial number         81B5074C1B6500076878
> cylinders             16383
> heads                 16
> sectors/track         63
> sector size           logical 512, physical 512, offset 0
> LBA supported         31277232 sectors
> LBA48 supported       31277232 sectors
> PIO supported         PIO4
> DMA supported         WDMA2 UDMA6
> media RPM             non-rotating
> Zoned-Device Commands no
> 
> Feature                      Support  Enabled   Value           Vendor
> read ahead                     yes      yes
> write cache                    yes      yes
> flush cache                    yes      yes
> overlap                        no
> Tagged Command Queuing (TCQ)   no       no
> Native Command Queuing (NCQ)   yes              32 tags
> NCQ Queue Management           no
> NCQ Streaming                  no
> Receive & Send FPDMA Queued    no
> SMART                          yes      yes
> microcode download             yes      yes
> security                       yes      no
> power management               yes      yes
> advanced power management      yes      no      0/0x00
> automatic acoustic management  no       no
> media status notification      no       no
> power-up in Standby            no       no
> write-read-verify              no       no
> unload                         yes      yes
> general purpose logging        yes      yes
> free-fall                      no       no
> Data Set Management (DSM/TRIM) yes
> DSM - max 512byte blocks       yes              8
> DSM - deterministic read       no
> Host Protected Area (HPA)      yes      no      31277232/31277232
> HPA - Security                 no
> 0#
> 
> 
> 
> On 8/21/2018 2:30 PM, Mike Tancsa wrote:
>> On 8/21/2018 9:51 AM, Eugene Grosbein wrote:
>>>
>>> It seems like faulty media to me: it silently returns bad data.
>>>
>>> There is an easy way to verify this just with naked eye:
>>>
>>> yes | dd bs=128k of=/dev/ada0
>>> hd /dev/ada0
>>>
>>> That is, hd(1) should write back only 3 lines of output:
>>>
>>> 00000000  79 0a 79 0a 79 0a 79 0a  79 0a 79 0a 79 0a 79 0a  
>>> |y.y.y.y.y.y.y.y.|
>>> *
>>> 01000000
>>>
>>> If not, the media if faulty.
>>>
>>
>> There are 3 of these disks I found. Unfortunately, they all seem a
>> little different from the revision stamps on the board.  They are all
>> from PCEngines who generally seem to source quality products. This is in
>> an APU3
>>
>> A "bad" disk
>>
>> 0# camcontrol identify ada0
>> pass0: <SATA SSD S9FM02.0> ACS-3 ATA SATA 3.x device
>> pass0: 600.000MB/s transfers (SATA 3.x, PIO4, PIO 8192bytes)
>>
>> protocol              ATA/ATAPI-10 SATA 3.x
>> device model          SATA SSD
>> firmware revision     S9FM02.0
>> serial number         DED9075313EC01677930
>> cylinders             16383
>> heads                 16
>> sectors/track         63
>> sector size           logical 512, physical 512, offset 0
>> LBA supported         31277232 sectors
>> LBA48 supported       31277232 sectors
>> PIO supported         PIO4
>> DMA supported         WDMA2 UDMA6
>> media RPM             non-rotating
>> Zoned-Device Commands no
>>
>> Feature                      Support  Enabled   Value           Vendor
>> read ahead                     yes      yes
>> write cache                    yes      yes
>> flush cache                    yes      yes
>> overlap                        no
>> Tagged Command Queuing (TCQ)   no       no
>> Native Command Queuing (NCQ)   yes              32 tags
>> NCQ Queue Management           no
>> NCQ Streaming                  no
>> Receive & Send FPDMA Queued    no
>> SMART                          yes      yes
>> microcode download             yes      yes
>> security                       yes      no
>> power management               yes      yes
>> advanced power management      yes      no      0/0x00
>> automatic acoustic management  no       no
>> media status notification      no       no
>> power-up in Standby            no       no
>> write-read-verify              no       no
>> unload                         yes      yes
>> general purpose logging        yes      yes
>> free-fall                      no       no
>> Data Set Management (DSM/TRIM) yes
>> DSM - max 512byte blocks       yes              8
>> DSM - deterministic read       no
>> Host Protected Area (HPA)      yes      no      31277232/31277232
>> HPA - Security                 no
>> 0#
>>
>> vs
>> a 'good' disk
>>
>> # camcontrol identify ada0
>> pass0: <SATA SSD SBFM01.0> ACS-4 ATA SATA 3.x device
>> pass0: 600.000MB/s transfers (SATA 3.x, PIO4, PIO 8192bytes)
>>
>> protocol              ATA/ATAPI-11 SATA 3.x
>> device model          SATA SSD
>> firmware revision     SBFM01.0
>> serial number         A44907781CE300040613
>> WWN                   5000000000000000
>> cylinders             16383
>> heads                 16
>> sectors/track         63
>> sector size           logical 512, physical 512, offset 0
>> LBA supported         31277232 sectors
>> LBA48 supported       31277232 sectors
>> PIO supported         PIO4
>> DMA supported         WDMA2 UDMA6
>> media RPM             non-rotating
>> Zoned-Device Commands no
>>
>> Feature                      Support  Enabled   Value           Vendor
>> read ahead                     yes      yes
>> write cache                    yes      yes
>> flush cache                    yes      yes
>> overlap                        no
>> Tagged Command Queuing (TCQ)   no       no
>> Native Command Queuing (NCQ)   yes              32 tags
>> NCQ Queue Management           no
>> NCQ Streaming                  no
>> Receive & Send FPDMA Queued    no
>> SMART                          yes      yes
>> microcode download             yes      yes
>> security                       yes      no
>> power management               yes      yes
>> advanced power management      no       no
>> automatic acoustic management  no       no
>> media status notification      no       no
>> power-up in Standby            no       no
>> write-read-verify              no       no
>> unload                         no       no
>> general purpose logging        yes      yes
>> free-fall                      no       no
>> Data Set Management (DSM/TRIM) yes
>> DSM - max 512byte blocks       yes              8
>> DSM - deterministic read       no
>> Host Protected Area (HPA)      yes      no      31277232/31277232
>> HPA - Security                 no
>>
>>
>>
>> % diff good bad
>> 1c1
>> < pass0: <SATA SSD SBFM01.0> ACS-4 ATA SATA 3.x device
>> ---
>>> pass0: <SATA SSD S9FM02.0> ACS-3 ATA SATA 3.x device
>> 4c4
>> < protocol              ATA/ATAPI-11 SATA 3.x
>> ---
>>> protocol              ATA/ATAPI-10 SATA 3.x
>> 6,8c6,7
>> < firmware revision     SBFM01.0
>> < serial number         A44907781CE300040613
>> < WWN                   5000000000000000
>> ---
>>> firmware revision     S9FM02.0
>>> serial number         DED9075313EC01677930
>> 34c33
>> < advanced power management      no       no
>> ---
>>> advanced power management      yes      no      0/0x00
>> 39c38
>> < unload                         no       no
>> ---
>>> unload                         yes      yes
>> 46a46
>>>
>>
>> 1# yes | dd bs=128k of=/dev/ada0
>> dd: /dev/ada0: short write on character device
>> dd: /dev/ada0: end of device
>> 0+1066621 records in
>> 122176+1 records out
>> 16013942784 bytes transferred in 566.461990 secs (28270110 bytes/sec)
>> 1# hd /dev/ada0
>> 00000000  79 0a 79 0a 79 0a 79 0a  79 0a 79 0a 79 0a 79 0a
>> |y.y.y.y.y.y.y.y.|
>> *
>> 3ba816000
>> 0#
>>
> 
> 


-- 
-------------------
Mike Tancsa, tel +1 519 651 3400 x203
Sentex Communications, m...@sentex.net
Providing Internet services since 1994 www.sentex.net
Cambridge, Ontario Canada
_______________________________________________
freebsd-stable@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"

Reply via email to