[U-Boot] [PATCH] Initilize IDE prior environment

[U-Boot] [PATCH] Initilize IDE prior environment

[Otavio Salvador]

To allow the loading of environment from IDE (for example, using FAT)
the initilization of IDE subsystem must come before the environment
one.

Successfully tested on QEMU x86.

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
---

 common/board_r.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/common/board_r.c b/common/board_r.c
index adc1f1937e..247a6668fc 100644
--- a/common/board_r.c
+++ b/common/board_r.c
@@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
 #ifdef CONFIG_MMC
 	initr_mmc,
 #endif
+#if defined(CONFIG_IDE)
+	initr_ide,
+#endif
 #ifdef CONFIG_HAS_DATAFLASH
 	initr_dataflash,
 #endif
@@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
 #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
 	initr_pcmcia,
 #endif
-#if defined(CONFIG_IDE)
-	initr_ide,
-#endif
 #ifdef CONFIG_LAST_STAGE_INIT
 	INIT_FUNC_WATCHDOG_RESET
 	/*
-- 
2.13.1

[U-Boot] Initilize IDE prior environment

[Tom Rini]

On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:

> To allow the loading of environment from IDE (for example, using FAT)
> the initilization of IDE subsystem must come before the environment
> one.
> 
> Successfully tested on QEMU x86.
> 
> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
> ---
> 
>  common/board_r.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/common/board_r.c b/common/board_r.c
> index adc1f1937e..247a6668fc 100644
> --- a/common/board_r.c
> +++ b/common/board_r.c
> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
>  #ifdef CONFIG_MMC
>  	initr_mmc,
>  #endif
> +#if defined(CONFIG_IDE)
> +	initr_ide,
> +#endif
>  #ifdef CONFIG_HAS_DATAFLASH
>  	initr_dataflash,
>  #endif
> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
>  #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
>  	initr_pcmcia,
>  #endif
> -#if defined(CONFIG_IDE)
> -	initr_ide,
> -#endif
>  #ifdef CONFIG_LAST_STAGE_INIT
>  	INIT_FUNC_WATCHDOG_RESET
>  	/*

Looking at the gap between these two areas, I see that we would
initialize PCI between the two areas.  So there's probably boards with
PCI IDE devices that'll regress if I take this now.  What's the answer?
Long term, device model.  Short term?  Is there a specific upstream
platform that you're seeing the problem or, or just custom / out of
tree?

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170620/d7b14a5c/attachment.sig>

[U-Boot] Initilize IDE prior environment

[Otavio Salvador]

On Tue, Jun 20, 2017 at 2:42 PM, Tom Rini <trini@konsulko.com> wrote:
> On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:
>
>> To allow the loading of environment from IDE (for example, using FAT)
>> the initilization of IDE subsystem must come before the environment
>> one.
>>
>> Successfully tested on QEMU x86.
>>
>> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
>> ---
>>
>>  common/board_r.c | 6 +++---
>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/common/board_r.c b/common/board_r.c
>> index adc1f1937e..247a6668fc 100644
>> --- a/common/board_r.c
>> +++ b/common/board_r.c
>> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
>>  #ifdef CONFIG_MMC
>>       initr_mmc,
>>  #endif
>> +#if defined(CONFIG_IDE)
>> +     initr_ide,
>> +#endif
>>  #ifdef CONFIG_HAS_DATAFLASH
>>       initr_dataflash,
>>  #endif
>> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
>>  #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
>>       initr_pcmcia,
>>  #endif
>> -#if defined(CONFIG_IDE)
>> -     initr_ide,
>> -#endif
>>  #ifdef CONFIG_LAST_STAGE_INIT
>>       INIT_FUNC_WATCHDOG_RESET
>>       /*
>
> Looking at the gap between these two areas, I see that we would
> initialize PCI between the two areas.  So there's probably boards with
> PCI IDE devices that'll regress if I take this now.  What's the answer?
> Long term, device model.  Short term?  Is there a specific upstream
> platform that you're seeing the problem or, or just custom / out of
> tree?

I am using this with QEMU. I enabled environment inside FAT and it
needs the IDE fix so the environment is properly loaded.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750

[U-Boot] Initilize IDE prior environment

[Bin Meng]

+Simon

On Wed, Jun 21, 2017 at 4:08 AM, Otavio Salvador
<otavio.salvador@ossystems.com.br> wrote:
> On Tue, Jun 20, 2017 at 2:42 PM, Tom Rini <trini@konsulko.com> wrote:
>> On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:
>>
>>> To allow the loading of environment from IDE (for example, using FAT)
>>> the initilization of IDE subsystem must come before the environment
>>> one.
>>>
>>> Successfully tested on QEMU x86.
>>>
>>> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
>>> ---
>>>
>>>  common/board_r.c | 6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/common/board_r.c b/common/board_r.c
>>> index adc1f1937e..247a6668fc 100644
>>> --- a/common/board_r.c
>>> +++ b/common/board_r.c
>>> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
>>>  #ifdef CONFIG_MMC
>>>       initr_mmc,
>>>  #endif
>>> +#if defined(CONFIG_IDE)
>>> +     initr_ide,
>>> +#endif
>>>  #ifdef CONFIG_HAS_DATAFLASH
>>>       initr_dataflash,
>>>  #endif
>>> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
>>>  #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
>>>       initr_pcmcia,
>>>  #endif
>>> -#if defined(CONFIG_IDE)
>>> -     initr_ide,
>>> -#endif
>>>  #ifdef CONFIG_LAST_STAGE_INIT
>>>       INIT_FUNC_WATCHDOG_RESET
>>>       /*
>>
>> Looking at the gap between these two areas, I see that we would
>> initialize PCI between the two areas.  So there's probably boards with
>> PCI IDE devices that'll regress if I take this now.  What's the answer?
>> Long term, device model.  Short term?  Is there a specific upstream
>> platform that you're seeing the problem or, or just custom / out of
>> tree?

I agree with Tom. Eventually we need convert IDE driver to driver model.

>
> I am using this with QEMU. I enabled environment inside FAT and it
> needs the IDE fix so the environment is properly loaded.
>

With driver model, IDE driver can be probed when U-Boot tries to load
environment from IDE.

For short term, not sure if we can accept this, Simon?

Regards,
Bin

[U-Boot] Initilize IDE prior environment

[Simon Glass]

Hi Bin,

On 20 June 2017 at 17:19, Bin Meng <bmeng.cn@gmail.com> wrote:
>
> +Simon
>
> On Wed, Jun 21, 2017 at 4:08 AM, Otavio Salvador
> <otavio.salvador@ossystems.com.br> wrote:
> > On Tue, Jun 20, 2017 at 2:42 PM, Tom Rini <trini@konsulko.com> wrote:
> >> On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:
> >>
> >>> To allow the loading of environment from IDE (for example, using FAT)
> >>> the initilization of IDE subsystem must come before the environment
> >>> one.
> >>>
> >>> Successfully tested on QEMU x86.
> >>>
> >>> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
> >>> ---
> >>>
> >>>  common/board_r.c | 6 +++---
> >>>  1 file changed, 3 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/common/board_r.c b/common/board_r.c
> >>> index adc1f1937e..247a6668fc 100644
> >>> --- a/common/board_r.c
> >>> +++ b/common/board_r.c
> >>> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
> >>>  #ifdef CONFIG_MMC
> >>>       initr_mmc,
> >>>  #endif
> >>> +#if defined(CONFIG_IDE)
> >>> +     initr_ide,
> >>> +#endif
> >>>  #ifdef CONFIG_HAS_DATAFLASH
> >>>       initr_dataflash,
> >>>  #endif
> >>> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
> >>>  #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
> >>>       initr_pcmcia,
> >>>  #endif
> >>> -#if defined(CONFIG_IDE)
> >>> -     initr_ide,
> >>> -#endif
> >>>  #ifdef CONFIG_LAST_STAGE_INIT
> >>>       INIT_FUNC_WATCHDOG_RESET
> >>>       /*
> >>
> >> Looking at the gap between these two areas, I see that we would
> >> initialize PCI between the two areas.  So there's probably boards with
> >> PCI IDE devices that'll regress if I take this now.  What's the answer?
> >> Long term, device model.  Short term?  Is there a specific upstream
> >> platform that you're seeing the problem or, or just custom / out of
> >> tree?
>
> I agree with Tom. Eventually we need convert IDE driver to driver model.
>
> >
> > I am using this with QEMU. I enabled environment inside FAT and it
> > needs the IDE fix so the environment is properly loaded.
> >
>
> With driver model, IDE driver can be probed when U-Boot tries to load
> environment from IDE.
>
> For short term, not sure if we can accept this, Simon?

If it does not break qemu and any platforms that use IDE (without PCI)
then I think it is fine.

It probably is not much work to convert IDE to driver model. Less if
we can remove more users :-)

Regards,
Simon

[U-Boot] Initilize IDE prior environment

[Tom Rini]

On Thu, Jun 22, 2017 at 10:05:48AM -0600, Simon Glass wrote:
> Hi Bin,
> 
> On 20 June 2017 at 17:19, Bin Meng <bmeng.cn@gmail.com> wrote:
> >
> > +Simon
> >
> > On Wed, Jun 21, 2017 at 4:08 AM, Otavio Salvador
> > <otavio.salvador@ossystems.com.br> wrote:
> > > On Tue, Jun 20, 2017 at 2:42 PM, Tom Rini <trini@konsulko.com> wrote:
> > >> On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:
> > >>
> > >>> To allow the loading of environment from IDE (for example, using FAT)
> > >>> the initilization of IDE subsystem must come before the environment
> > >>> one.
> > >>>
> > >>> Successfully tested on QEMU x86.
> > >>>
> > >>> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
> > >>> ---
> > >>>
> > >>>  common/board_r.c | 6 +++---
> > >>>  1 file changed, 3 insertions(+), 3 deletions(-)
> > >>>
> > >>> diff --git a/common/board_r.c b/common/board_r.c
> > >>> index adc1f1937e..247a6668fc 100644
> > >>> --- a/common/board_r.c
> > >>> +++ b/common/board_r.c
> > >>> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
> > >>>  #ifdef CONFIG_MMC
> > >>>       initr_mmc,
> > >>>  #endif
> > >>> +#if defined(CONFIG_IDE)
> > >>> +     initr_ide,
> > >>> +#endif
> > >>>  #ifdef CONFIG_HAS_DATAFLASH
> > >>>       initr_dataflash,
> > >>>  #endif
> > >>> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
> > >>>  #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
> > >>>       initr_pcmcia,
> > >>>  #endif
> > >>> -#if defined(CONFIG_IDE)
> > >>> -     initr_ide,
> > >>> -#endif
> > >>>  #ifdef CONFIG_LAST_STAGE_INIT
> > >>>       INIT_FUNC_WATCHDOG_RESET
> > >>>       /*
> > >>
> > >> Looking at the gap between these two areas, I see that we would
> > >> initialize PCI between the two areas.  So there's probably boards with
> > >> PCI IDE devices that'll regress if I take this now.  What's the answer?
> > >> Long term, device model.  Short term?  Is there a specific upstream
> > >> platform that you're seeing the problem or, or just custom / out of
> > >> tree?
> >
> > I agree with Tom. Eventually we need convert IDE driver to driver model.
> >
> > >
> > > I am using this with QEMU. I enabled environment inside FAT and it
> > > needs the IDE fix so the environment is properly loaded.
> > >
> >
> > With driver model, IDE driver can be probed when U-Boot tries to load
> > environment from IDE.
> >
> > For short term, not sure if we can accept this, Simon?
> 
> If it does not break qemu and any platforms that use IDE (without PCI)
> then I think it is fine.
> 
> It probably is not much work to convert IDE to driver model. Less if
> we can remove more users :-)

So, I see configs with both CONFIG_IDE and CONFIG_PCI set, so I'm not
going to take this patch.  We can move env on qemu easily enough I
think.  Sorry.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20170622/5f4d81f9/attachment.sig>

[U-Boot] Initilize IDE prior environment

[Heiko Schocher]

Hello,

Am 23.06.2017 um 01:49 schrieb Tom Rini:
> On Thu, Jun 22, 2017 at 10:05:48AM -0600, Simon Glass wrote:
>> Hi Bin,
>>
>> On 20 June 2017 at 17:19, Bin Meng <bmeng.cn@gmail.com> wrote:
>>>
>>> +Simon
>>>
>>> On Wed, Jun 21, 2017 at 4:08 AM, Otavio Salvador
>>> <otavio.salvador@ossystems.com.br> wrote:
>>>> On Tue, Jun 20, 2017 at 2:42 PM, Tom Rini <trini@konsulko.com> wrote:
>>>>> On Mon, Jun 19, 2017 at 11:56:04AM -0300, Otavio Salvador wrote:
>>>>>
>>>>>> To allow the loading of environment from IDE (for example, using FAT)
>>>>>> the initilization of IDE subsystem must come before the environment
>>>>>> one.
>>>>>>
>>>>>> Successfully tested on QEMU x86.
>>>>>>
>>>>>> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
>>>>>> ---
>>>>>>
>>>>>>   common/board_r.c | 6 +++---
>>>>>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>>>>>
>>>>>> diff --git a/common/board_r.c b/common/board_r.c
>>>>>> index adc1f1937e..247a6668fc 100644
>>>>>> --- a/common/board_r.c
>>>>>> +++ b/common/board_r.c
>>>>>> @@ -806,6 +806,9 @@ static init_fnc_t init_sequence_r[] = {
>>>>>>   #ifdef CONFIG_MMC
>>>>>>        initr_mmc,
>>>>>>   #endif
>>>>>> +#if defined(CONFIG_IDE)
>>>>>> +     initr_ide,
>>>>>> +#endif
>>>>>>   #ifdef CONFIG_HAS_DATAFLASH
>>>>>>        initr_dataflash,
>>>>>>   #endif
>>>>>> @@ -878,9 +881,6 @@ static init_fnc_t init_sequence_r[] = {
>>>>>>   #if defined(CONFIG_CMD_PCMCIA) && !defined(CONFIG_IDE)
>>>>>>        initr_pcmcia,
>>>>>>   #endif
>>>>>> -#if defined(CONFIG_IDE)
>>>>>> -     initr_ide,
>>>>>> -#endif
>>>>>>   #ifdef CONFIG_LAST_STAGE_INIT
>>>>>>        INIT_FUNC_WATCHDOG_RESET
>>>>>>        /*
>>>>>
>>>>> Looking at the gap between these two areas, I see that we would
>>>>> initialize PCI between the two areas.  So there's probably boards with
>>>>> PCI IDE devices that'll regress if I take this now.  What's the answer?
>>>>> Long term, device model.  Short term?  Is there a specific upstream
>>>>> platform that you're seeing the problem or, or just custom / out of
>>>>> tree?
>>>
>>> I agree with Tom. Eventually we need convert IDE driver to driver model.
>>>
>>>>
>>>> I am using this with QEMU. I enabled environment inside FAT and it
>>>> needs the IDE fix so the environment is properly loaded.
>>>>
>>>
>>> With driver model, IDE driver can be probed when U-Boot tries to load
>>> environment from IDE.

Yup, that would be the best.

>>> For short term, not sure if we can accept this, Simon?
>>
>> If it does not break qemu and any platforms that use IDE (without PCI)
>> then I think it is fine.
>>
>> It probably is not much work to convert IDE to driver model. Less if
>> we can remove more users :-)
>
> So, I see configs with both CONFIG_IDE and CONFIG_PCI set, so I'm not
> going to take this patch.  We can move env on qemu easily enough I
> think.  Sorry.

May a short workaround ...
Why not loading an env.txt over ide after U-Boot is up and import it?

bye,
Heiko
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany