Debian on HP T5325 Thin Client

@CV & @xssa - thank you for the images and details of how you did this.

Xssa,

Wow! Nice works :)

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

bodhi,

Is it possible to setup ledtrig-ide-disk to monitor not only SATA1 activity but both SATA1 and SATA2?
For now green led blinks only on SATA1.

Thank You.

Hi xssa,

The SATA 2nd port LED trigger is already supported in the kernel with my patch for the SATA driver. But the T5325 has only one exposed SATA. Now that you and CV have the 2nd SATA :) However, it seems that the T5325 SATA LED is hardwired (I did not need to do anything in the DTS for it to blink).

It is good that you've verified that hardwired only to the SATA port 0! I don't know yet if it is good or bad news, but I'll do some investigation to see if there is anyway to activate it.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@all - syong has ported the kernel driver for the frame buffer if anyone is willing to test, he would appreciate feedback in the graphics section.

http://forum.doozan.com/read.php?9,26555,27952,page=4#msg-27952

I would test but im not near my "toys" for a few days



Edited 1 time(s). Last edit at 05/09/2016 05:04PM by Gravelrash.

Gravelrash Wrote:
-------------------------------------------------------
> @all - syong has ported the kernel driver for the
> frame buffer if anyone is willing to test, he
> would appreciate feedback in the graphics
> section.
>
> http://forum.doozan.com/read.php?9,26555,27952,pag
> e=4#msg-27952
>
> I would test but im not near my "toys" for a few
> days

I did not see syong upload any driver?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@bodhi

its me in a dolt - i can only blame late nights and foreign diets :o) - i should have said XSSA posted it. its here

http://forum.doozan.com/read.php?9,26555,page=4

syong Wrote:
-------------------------------------------------------
> fw_printenv and fw_setenv work on stock uBoot if
> the fw_env.config is set as follows:
>
>

> (17:52)(hep)~> cat /etc/fw_env.config
> /dev/mtd2 0 0x1000
>

>
> where mtd2 could be mtd3 or mtd4 for the other
> environments.
>

Correction -- this works only for t5325 with SST25VF080B flash.

-syong

I need some help figuring out the SPI flash, if anybody can boot into stock OS. Look for this binary

cd /
find . -name 'flashcp'

And please post the binary if you found it.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

bodhi,

If you looked at the source of flashcp, I think you would rather use dd instead.

-syong

syong Wrote:
-------------------------------------------------------
> bodhi,
>
> If you looked at the source of flashcp, I think
> you would rather use dd instead.

Really, it's that bad? :)

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@syong,

The reason I want to see the flashcp binary fromstock OS is: when you look at the HP recovery USB, they restore stock u-boot using flashcp. So I just want to test that version.

Linux kernel 4.5 seems to take care of the SPI flash correctly (it detects the addressing mode of the flash in the driver, ie. 3-byte or 4-byte). So flashing it should work. But somehow we can't. The error is

Verifying data: 10k/0k (64%)File does not seem to match flash data. First mismatch at 0x00000000-0x00002800

This is the addressing mode error, AFAICT.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@bodhi,

Have you looked at the source of flashcp binary from stock OS yet?

-syong



Edited 1 time(s). Last edit at 05/24/2016 09:16AM by syong.

syong Wrote:
-------------------------------------------------------
> @bodhi,
>
> Have you looked at the source of flashcp binary
> from stock OS yet?

Syong,

Not yet. Is it in the GPL source tree?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@bodhi,

Hacking without looking at the source will be quite difficult.

-syong



Edited 1 time(s). Last edit at 05/24/2016 07:43PM by syong.

syong Wrote:
-------------------------------------------------------
> @bodhi,
>
> Hacking without looking at the source will be
> quite difficult.

:)) I know. Just being lazy because I have a hunch that old flashcp running in stock OS will work (address mode was hard coded). And dd is not a good way to flash so I'd like to avoid it.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Can state that SPI reading on the last kernels are broken - data taken from mtdX partitions does not match original at all. So managing uboot envs is impossible on last kernels. I'm running stock uboot, kernels 4.4.0, 4.5.0, 4.5.2. SPI flash is mx25l8005 (from dmesg)

xssa Wrote:
-------------------------------------------------------
> Can state that SPI reading on the last kernels are
> broken - data taken from mtdX partitions does not
> match original at all. So managing uboot envs is
> impossible on last kernels. I'm running stock
> uboot, kernels 4.4.0, 4.5.0, 4.5.2. SPI flash is
> mx25l8005 (from dmesg)

If you are running stock u-boot, you should not try to change u-boot envs in Debian. This is most likey has nothing to do with the fact that it is SPI .

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

bodhi Wrote:
-------------------------------------------------------
> If you are running stock u-boot, you should not
> try to change u-boot envs in Debian. This is most
> likey has nothing to do with the fact that it is
> SPI .
I'm fine with that. Did you try to figure out where the problem is? I think it is on the kernel side. Both cat and dd from /dev/mtdX or /dev/mtdblockX gives same broken data.
For now i'm focusing on porting xga accelerated driver into recent xorg-server on my spare time.

xssa,

> I'm fine with that. Did you try to figure out
> where the problem is?

problem with the envs or the problem with flashing using flashcp?

> I think it is on the kernel
> side. Both cat and dd from /dev/mtdX or
> /dev/mtdblockX gives same broken data.

If you meant the envs then, of course it will likely not give meaningful data. The checksums are different. The kernel uses newer-u-boot-type of checksum.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@xxsa:

Since the problem of SPI reading is broken with MX25L8005 and Numonyx-25P80 flashes, but not on SST25VF080B flash, I agree with xssa that it is on the kernel side.

@bodhi:

Forget about the env and flashcp.c. We should read m25p80.c now.

-syong

As i was working with sst25vfxxx SPI flash and other SPIs compatible with 25p80 generic driver in linux few years ago, i can state they use different determination and access methods. Anyone can google datasheets and find differences.
Somebody knows last kernel can read mx25l8005 w/o errors?



Edited 1 time(s). Last edit at 05/26/2016 01:11PM by xssa.

syong,

Are you running u-boot 2015.10 and new kernel 4.6 on the box with SST25VF080B?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@everybody,
I'd like to know if you run the new u-boot 2015.10 with the latest kernel 4.5 or 4.6, which SPI flash version is not working.

@syong,
Note that I've said stock u-boot and new kernel SPI MTDs don't work together. The mp25p80 looks fine last time I've read it.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 05/26/2016 07:47PM by bodhi.

@xxsa:

I just checked again and here is the findings:

1. With mx25l8005, reading is broken.
2. With numonyx-25p80, reading is ok, but writing is broken.
3. With sst25vf080b, both reading and writing are ok.

I am also interested to know if somebody can read mx25l8005 w/o errors with recent kernel.

@bodhi:

> Are you running u-boot 2015.10 and new kernel 4.6 on the box with SST25VF080B?

No, I am still on the stock u-boot. Just too paranoid to flash a new u-boot without looking at its source first.:)

> I'd like to know if you run the new u-boot 2015.10 with the latest kernel 4.5 or 4.6, which SPI flash version is not working.

What SPI flash version is in your box? If you suspect the addressing mode error, have you looked at spi-nor.c and tried to hard code to use either 3-byte or 4-byte?

> Note that I've said stock u-boot and new kernel SPI MTDs don't work together. The mp25p80 looks fine last time I've read it.

But I have said stock u-boot and new kernel SPI MTDs work together just fine for box with sst25vf080b. Regarding m25p80.c, spi-nor.c looks more relevant.

-syong

root@t5325:~# dmesg |grep spi
[    1.215029] m25p80 spi0.0: found mx25l8005, expected m25p80
[    1.220756] m25p80 spi0.0: mx25l8005 (1024 Kbytes)
[    1.225619] 5 ofpart partitions found on MTD device spi0.0

@syong - can you look to dmesq and find what driver is used to deal with numonix-25p80 and sst25vf080b?



Edited 2 time(s). Last edit at 05/26/2016 10:55PM by xssa.

syong,

> What SPI flash version is in your box? If you suspect the addressing mode error, have you looked at spi-nor.c and tried to hard code to use either 3-byte or 4-byte?

I don't think this is needed. The kernel driver takes care of 3,4 bytes addr already, if I understood it correctly.

> No, I am still on the stock u-boot. Just too
> paranoid to flash a new u-boot without looking at
> its source first.:)

Ok. A little paranoid is always good! when you're not paranoid at all is when you got into trouble :) that's what I always remind my younger relatives.

You don't have to flash new u-boot, just run it with kwboot, like CV, boba, Leggo, and Gravelrash did previously in this thread. That would be a really good test to see if all 3 types of SPI flash in this box will work with the new u-boot and new kernel.

Should not do it the hard way if you're not going to keep stock u-boot in the future.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 2 time(s). Last edit at 05/26/2016 11:43PM by bodhi.

@xssa:

All three flashes used the m25p80 driver.

@bodhi:

Can you explain why stock u-boot and new kernel SPI MTDs work together just fine for box with sst25vf080b?

-syong

syong,

> Can you explain why stock u-boot and new kernel
> SPI MTDs work together just fine for box with
> sst25vf080b?

Did you try kwboot this combination on sst25vf080b ?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 05/27/2016 03:20PM by bodhi.

@Gravelrash,

> Actually joking aside (cos im still nowhere near
> my toys and i have a backlog of feedback i
> promised to do - sorry @bodhi et al).

No problem Gravelrash :) when you're back I'll need to get some information from stock OS. I believe you still have one running pure stock.

I'm trying to figure out why SSH does not work with stock OS running (searching this rootfs did not turn up anything useful). There must be some settings in stock rootfs that disable SSH root login. I don't recall there is another user on this that you can log in with SSH?

ps -deaf

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)