Skip to contentSkip to conversation

Regression in U-boot v2019-07-rcX

U-Boot Andy Shevchenko 4 participants 8 messages Jun 12, 2019 – Jun 17, 2019

lore
Share & tools
lore
#0 lore raw

Hi!

I have noticed today the regression in newer U-Boot.
For now I'm not sure which version is affected first. At least v2019.07-rc4 behaves badly.

Under badly I mean the following. Consider the output at the boot time:

U-Boot 2019.07-rc4-00012-gcea7942d57 (Jun 12 2019 - 16:08:09 +0300)

CPU:   Intel(R) Atom(TM) CPU  U1000  @  500MHz
DRAM:  980.6 MiB
MMC:   mmc at ff3fc000: 0, mmc at ff3fa000: 1
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   No ethernet found.
Hit any key to stop autoboot:  0 

Note lines started with DRAM: and MMC:. Now the 4 seconds (!) time is passed between them.

On top of that DFU seems twice slower (on the glance, I haven't measure with time yet).

Any ideas, while I'll try to allocate some time to investigate myself?

P.S. The board I'm testing with is Intel Edison. I think no need to tell that nothing has been changed except the base (I'm rebased my stuff on top of latest and greatest U-Boot releases).

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#1 lore raw
Quoted 23 lines · Andy Shevchenko

Hi!

I have noticed today the regression in newer U-Boot.
For now I'm not sure which version is affected first. At least v2019.07-rc4 behaves badly.

Under badly I mean the following. Consider the output at the boot time:

U-Boot 2019.07-rc4-00012-gcea7942d57 (Jun 12 2019 - 16:08:09 +0300)

CPU:   Intel(R) Atom(TM) CPU  U1000  @  500MHz
DRAM:  980.6 MiB
MMC:   mmc at ff3fc000: 0, mmc at ff3fa000: 1
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   No ethernet found.
Hit any key to stop autoboot:  0 

Note lines started with DRAM: and MMC:. Now the 4 seconds (!) time is passed between them.

Quoted 8 lines

On top of that DFU seems twice slower (on the glance, I haven't measure with time yet).

Any ideas, while I'll try to allocate some time to investigate myself?

P.S. The board I'm testing with is Intel Edison. I think no need to tell that nothing has been changed except the base (I'm rebased my stuff on top of latest and greatest U-Boot releases).

v2019.07-rc1 is NOT affected!

The time difference with DFU:
-rc1: real 0m1.114s
-rc4: real 0m2.691s

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#2 lore raw

Hi Andy,

Quoted 39 lines · Andy Shevchenko
Quoted 27 lines

Hi!

I have noticed today the regression in newer U-Boot. For now I'm not sure which version is affected first. At least v2019.07-rc4 behaves badly.

Under badly I mean the following. Consider the output at the boot time:

U-Boot 2019.07-rc4-00012-gcea7942d57 (Jun 12 2019 - 16:08:09 +0300)

CPU:   Intel(R) Atom(TM) CPU  U1000  @  500MHz
DRAM:  980.6 MiB
MMC:   mmc at ff3fc000: 0, mmc at ff3fa000: 1
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   No ethernet found.
Hit any key to stop autoboot:  0 

Note lines started with DRAM: and MMC:. Now the 4 seconds (!) time is passed between them.

On top of that DFU seems twice slower (on the glance, I haven't measure with time yet).

Quoted 6 lines

Any ideas, while I'll try to allocate some time to investigate myself?

P.S. The board I'm testing with is Intel Edison. I think no need to tell that nothing has been changed except the base (I'm rebased my stuff on top of latest and greatest U-Boot releases).

v2019.07-rc1 is NOT affected!

The time difference with DFU:

Review

Quoted 2 lines
(quoted diff) +0 -2
-rc1: real    0m1.114s
-rc4: real    0m2.691s

Could you bisect between -rc1 and -rc4 ?

Thanks in advance,

Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma at denx.de -------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20190612/7fd25915/attachment.sig>

#4 lore raw
In reply to Andy Shevchenko
Quoted 36 lines · Andy Shevchenko
Quoted 23 lines · Andy Shevchenko

Hi!

I have noticed today the regression in newer U-Boot.
For now I'm not sure which version is affected first. At least v2019.07-rc4 behaves badly.

Under badly I mean the following. Consider the output at the boot time:

U-Boot 2019.07-rc4-00012-gcea7942d57 (Jun 12 2019 - 16:08:09 +0300)

CPU:   Intel(R) Atom(TM) CPU  U1000  @  500MHz
DRAM:  980.6 MiB
MMC:   mmc at ff3fc000: 0, mmc at ff3fa000: 1
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   No ethernet found.
Hit any key to stop autoboot:  0 

Note lines started with DRAM: and MMC:. Now the 4 seconds (!) time is passed between them.

Quoted 8 lines

On top of that DFU seems twice slower (on the glance, I haven't measure with time yet).

Any ideas, while I'll try to allocate some time to investigate myself?

P.S. The board I'm testing with is Intel Edison. I think no need to tell that nothing has been changed except the base (I'm rebased my stuff on top of latest and greatest U-Boot releases).

v2019.07-rc1 is NOT affected!

The time difference with DFU:

Review

Quoted 2 lines
(quoted diff) +0 -2
-rc1: real    0m1.114s
-rc4: real    0m2.691s

The culprit commit is:

commit 665cb18ea64aabbeb03d27a4c92ddec1baccb87a

Author: Simon Glass <[email protected]>
Date:   Thu Apr 25 21:59:06 2019 -0600

x86: Don't set up MTRRs in SPL

Please revert ASAP before release, thanks!

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#3 lore raw
In reply to Lukasz Majewski
Quoted 4 lines · Lukasz Majewski
Quoted 4 lines · Andy Shevchenko

v2019.07-rc1 is NOT affected!

Quoted 8 lines

Review

Quoted 4 lines

The time difference with DFU:

(quoted diff) +0 -2
-rc1: real    0m1.114s
-rc4: real    0m2.691s

Could you bisect between -rc1 and -rc4 ?

Thanks in advance,

Just did, see my previous message.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#5 lore raw
In reply to Andy Shevchenko
Quoted 44 lines · Andy Shevchenko
Quoted 36 lines · Andy Shevchenko
Quoted 23 lines · Andy Shevchenko

Hi!

I have noticed today the regression in newer U-Boot.
For now I'm not sure which version is affected first. At least v2019.07-rc4 behaves badly.

Under badly I mean the following. Consider the output at the boot time:

U-Boot 2019.07-rc4-00012-gcea7942d57 (Jun 12 2019 - 16:08:09 +0300)

CPU:   Intel(R) Atom(TM) CPU  U1000  @  500MHz
DRAM:  980.6 MiB
MMC:   mmc at ff3fc000: 0, mmc at ff3fa000: 1
Loading Environment from MMC... OK
In:    serial
Out:   serial
Err:   serial
Net:   No ethernet found.
Hit any key to stop autoboot:  0

Note lines started with DRAM: and MMC:. Now the 4 seconds (!) time is passed between them.

Quoted 8 lines

On top of that DFU seems twice slower (on the glance, I haven't measure with time yet).

Any ideas, while I'll try to allocate some time to investigate myself?

P.S. The board I'm testing with is Intel Edison. I think no need to tell that nothing has been changed except the base (I'm rebased my stuff on top of latest and greatest U-Boot releases).

v2019.07-rc1 is NOT affected!

The time difference with DFU:

Review

Quoted 2 lines
(quoted diff) +0 -2
-rc1: real    0m1.114s
-rc4: real    0m2.691s

The culprit commit is:

Quoted 8 lines

commit 665cb18ea64aabbeb03d27a4c92ddec1baccb87a

Author: Simon Glass <[email protected]>
Date:   Thu Apr 25 21:59:06 2019 -0600

x86: Don't set up MTRRs in SPL

Please revert ASAP before release, thanks!

So it looks that MTRRs are not programmed for Intel Edison to enable cache?

Simon, would you please take a look? I suspect simply revert this will break the Chromebook SPL build?

Regards,
Bin

#6 lore raw
Quoted 12 lines · Bin Meng
Quoted 10 lines · Andy Shevchenko
Quoted 1 line · Andy Shevchenko

On Wed, Jun 12, 2019 at 04:18:25PM +0300, Andy Shevchenko wrote:

commit 665cb18ea64aabbeb03d27a4c92ddec1baccb87a

Author: Simon Glass <[email protected]>
Date:   Thu Apr 25 21:59:06 2019 -0600

x86: Don't set up MTRRs in SPL

Quoted 6 lines
Quoted 1 line

Please revert ASAP before release, thanks!

So it looks that MTRRs are not programmed for Intel Edison to enable cache?

Simon, would you please take a look? I suspect simply revert this will break the Chromebook SPL build?

Since there is no activity on this and release is coming, I would propose to revert for now.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#7 lore raw

Hi Bin, Andy,

Quoted 16 lines · Andy Shevchenko
Quoted 14 lines · Bin Meng
Quoted 11 lines · Andy Shevchenko
Quoted 1 line · Andy Shevchenko

On Wed, Jun 12, 2019 at 04:18:25PM +0300, Andy Shevchenko wrote:

commit 665cb18ea64aabbeb03d27a4c92ddec1baccb87a

Author: Simon Glass <[email protected]>
Date:   Thu Apr 25 21:59:06 2019 -0600

x86: Don't set up MTRRs in SPL

Please revert ASAP before release, thanks!

Quoted 7 lines
Quoted 4 lines

So it looks that MTRRs are not programmed for Intel Edison to enable cache?

Simon, would you please take a look? I suspect simply revert this will break the Chromebook SPL build?

Since there is no activity on this and release is coming, I would propose to revert for now.

I am OK with a revert for now if we don't have another solution.

The problem here is that we want to select when the mtrrs are programmed.

With my patch, this code is enabled on edison. I think the solution might be to set the mtrrs earlier on link. But I'll need to take a look and it won't be for about two weeks.

Regards,
Simon