Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Burton, Ross]

On 17 November 2017 at 06:08, Hongxu Jia  wrote:

> +qemu by default emulates IDE and the linux-yocto kernel(s) use
> +CONFIG_IDE instead of the more modern libsata, so disks appear as
> +/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
> +is deprecated.
>

Can we fix our kernel and/or qemu configuration too?

Ross
-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

On 2017年11月20日 19:07, Burton, Ross wrote:
On 17 November 2017 at 06:08, Hongxu Jia > wrote:


+qemu by default emulates IDE and the linux-yocto kernel(s) use
+CONFIG_IDE instead of the more modern libsata, so disks appear as
+/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
+is deprecated.


Can we fix our kernel and/or qemu configuration too?



Hi Bruce,

Do you have any suggestion? How about disable CONFIG_IDE?

//Hongxu


Ross



-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

On 2017年11月20日 19:07, Burton, Ross wrote:
On 17 November 2017 at 06:08, Hongxu Jia > wrote:


+qemu by default emulates IDE and the linux-yocto kernel(s) use
+CONFIG_IDE instead of the more modern libsata, so disks appear as
+/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
+is deprecated.


Can we fix our kernel and/or qemu configuration too?



For qemu, RP have replace it with virtio, but it have the
similar issue on virtualbox with IDE controller.
...
commit f59e729f98ef9b506b0cfdc415567e03ec87f2a9
Author: Richard Purdie 
Date:   Tue Aug 22 22:58:02 2017 +0100

    runqemu: Use virtio to mount cdrom drives

    The IDE driver in the kernel is fragile and in 4.12 is causing 
backtraces.
    To unblock 4.12 kernel merging use the virtio CD driver instead to 
mount

    iso images which should be faster and more stable.
...

//Hongxu


Ross



-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Andre McCurdy]

On Mon, Nov 20, 2017 at 5:21 PM, Hongxu Jia  wrote:
> On 2017年11月20日 19:07, Burton, Ross wrote:
> On 17 November 2017 at 06:08, Hongxu Jia  wrote:
>>
>> +qemu by default emulates IDE and the linux-yocto kernel(s) use
>> +CONFIG_IDE instead of the more modern libsata, so disks appear as
>> +/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
>> +is deprecated.
>
> Can we fix our kernel and/or qemu configuration too?
>
> For qemu, RP have replace it with virtio, but it have the
> similar issue on virtualbox with IDE controller.

VirtualBox gives you the choice about whether to connect the HDD image
to a virtual IDE or SATA port.

ie VirtualBox isn't a reason to have our rootfs images continue to try
to support IDE.

> ...
> commit f59e729f98ef9b506b0cfdc415567e03ec87f2a9
> Author: Richard Purdie 
> Date:   Tue Aug 22 22:58:02 2017 +0100
>
> runqemu: Use virtio to mount cdrom drives
>
> The IDE driver in the kernel is fragile and in 4.12 is causing
> backtraces.
> To unblock 4.12 kernel merging use the virtio CD driver instead to mount
> iso images which should be faster and more stable.
> ...
>
> //Hongxu
>
> Ross
>
> --
> ___
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core
>
-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

On 2017年11月21日 10:00, Andre McCurdy wrote:

On Mon, Nov 20, 2017 at 5:21 PM, Hongxu Jia  wrote:

On 2017年11月20日 19:07, Burton, Ross wrote:
On 17 November 2017 at 06:08, Hongxu Jia  wrote:

+qemu by default emulates IDE and the linux-yocto kernel(s) use
+CONFIG_IDE instead of the more modern libsata, so disks appear as
+/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
+is deprecated.

Can we fix our kernel and/or qemu configuration too?

For qemu, RP have replace it with virtio, but it have the
similar issue on virtualbox with IDE controller.

VirtualBox gives you the choice about whether to connect the HDD image
to a virtual IDE or SATA port.

ie VirtualBox isn't a reason to have our rootfs images continue to try
to support IDE.


For IDE, the mainstream distribution (ubuntu/fedora) do not support it.

Ubuntu 17.10, kernel 4.13.0-16
vim /boot/config-4.13.0-16-generic
...
# CONFIG_IDE is not set
CONFIG_ATA_PIIX=y
...

Fedora 27, kernel 4.13.5
vi /boot/config-4.13.5-300.fc27.x86_64
...
# CONFIG_IDE is not set
CONFIG_ATA_PIIX=y
...

The IDE seems not stable, and I have fixed two issue on it.
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=79d73346ac05bc31f2e96f899c4e9616a8d4
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8dc7a31fbce5e2dbbacd83d910da37105181b054

If enable both of IDE and ATA and the IDE still have a higher priority.
We should explicitly not set CONFIG_IDE.

Here is the discussion on upstream mailing list.
https://www.mail-archive.com/linux-block@vger.kernel.org/msg15114.html

//Hongxu


...
commit f59e729f98ef9b506b0cfdc415567e03ec87f2a9
Author: Richard Purdie 
Date:   Tue Aug 22 22:58:02 2017 +0100

 runqemu: Use virtio to mount cdrom drives

 The IDE driver in the kernel is fragile and in 4.12 is causing
backtraces.
 To unblock 4.12 kernel merging use the virtio CD driver instead to mount
 iso images which should be faster and more stable.
...

//Hongxu

Ross

--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core



--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Bruce Ashfield]

On Mon, Nov 20, 2017 at 8:18 PM, Hongxu Jia 
wrote:

> On 2017年11月20日 19:07, Burton, Ross wrote:
>
> On 17 November 2017 at 06:08, Hongxu Jia  wrote:
>
>> +qemu by default emulates IDE and the linux-yocto kernel(s) use
>> +CONFIG_IDE instead of the more modern libsata, so disks appear as
>> +/dev/hd*. A similar patch rejected by upstream because CONFIG_IDE
>> +is deprecated.
>>
>
> Can we fix our kernel and/or qemu configuration too?
>
>
> Hi Bruce,
>
> Do you have any suggestion? How about disable CONFIG_IDE?
>

I've been telling people we can disable CONFIG_IDE for years now.

It just requires the kernel-cache change, and someone to boot all arches to
show that
it continues to work, as well as a quick search for references to hda to
make sure they
are all gone.

Not something that is on my TODO list at the moment, but I'd take patches
to the kernel
part (with that testing shown).

Bruce


>
> //Hongxu
>
> Ross
>
>
>
> --
> ___
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core
>
>


-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end"
-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

On 2017年11月21日 11:04, Bruce Ashfield wrote:



On Mon, Nov 20, 2017 at 8:18 PM, Hongxu Jia > wrote:


On 2017年11月20日 19:07, Burton, Ross wrote:

On 17 November 2017 at 06:08, Hongxu Jia
mailto:hongxu@windriver.com>> wrote:

+qemu by default emulates IDE and the linux-yocto kernel(s) use
+CONFIG_IDE instead of the more modern libsata, so disks
appear as
+/dev/hd*. A similar patch rejected by upstream because
CONFIG_IDE
+is deprecated.


Can we fix our kernel and/or qemu configuration too?



Hi Bruce,

Do you have any suggestion? How about disable CONFIG_IDE?


I've been telling people we can disable CONFIG_IDE for years now.

It just requires the kernel-cache change, and someone to boot all 
arches to show that
it continues to work, as well as a quick search for references to hda 
to make sure they

are all gone.

Not something that is on my TODO list at the moment, but I'd take 
patches to the kernel

part (with that testing shown).



I think I could have a try to clean up CONFIG_IDE.

Two steps:
1. Tweak kernel-cache to remove CONFIG_IDE for all available bsp.
    It require necessary test to prove it is safe to remove it.
    If the test is OK, I will send patch to yocto mailing list.

2. After 1 merged to kernel, I will clean up hda related (systemd/udev)
    in oe-core, and send the fix to this mailing list.

//Hongxu


Bruce


//Hongxu


Ross




--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org

http://lists.openembedded.org/mailman/listinfo/openembedded-core





--
"Thou shalt not follow the NULL pointer, for chaos and madness await 
thee at its end"



-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Burton, Ross]

On 21 November 2017 at 07:00, Hongxu Jia  wrote:

> 1. Tweak kernel-cache to remove CONFIG_IDE for all available bsp.
> It require necessary test to prove it is safe to remove it.
> If the test is OK, I will send patch to yocto mailing list.
>

I know you have your own huge autobuilder framework inside WR but I'll
happily throw a patch against oe-core at the public autobuilders for extra
testing.

Ross
-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

On 2017年11月21日 19:31, Burton, Ross wrote:
On 21 November 2017 at 07:00, Hongxu Jia > wrote:


1. Tweak kernel-cache to remove CONFIG_IDE for all available bsp.
    It require necessary test to prove it is safe to remove it.
    If the test is OK, I will send patch to yocto mailing list.


I know you have your own huge autobuilder framework inside WR but I'll 
happily throw a patch against oe-core at the public autobuilders for 
extra testing.




No problem, I would like the public autobuilders.

But there several banners:

1. I do not have account on public autobuilder
    https://autobuilder.yocto.io/

    Maybe I should ask Michael or Joshua for help

2. The fix is on kernel-cache, not directly on oe-core,
    So it is difficult to use autobuilder

   Maybe I should ask Bruce for help

3. Most importantly, how to test, when find a way,
    the ideally is to add the test case to oe-core's
    meta-selftest

   I should do some investigation on the storage test.

So I think I should do 3 first, and then 1/2 later

//Hongxu


Ross



-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Bruce Ashfield]

On Tue, Nov 21, 2017 at 10:36 PM, Hongxu Jia 
wrote:

> On 2017年11月21日 19:31, Burton, Ross wrote:
>
> On 21 November 2017 at 07:00, Hongxu Jia  wrote:
>
>> 1. Tweak kernel-cache to remove CONFIG_IDE for all available bsp.
>> It require necessary test to prove it is safe to remove it.
>> If the test is OK, I will send patch to yocto mailing list.
>>
>
> I know you have your own huge autobuilder framework inside WR but I'll
> happily throw a patch against oe-core at the public autobuilders for extra
> testing.
>
>
> No problem, I would like the public autobuilders.
>
> But there several banners:
>
> 1. I do not have account on public autobuilder
> https://autobuilder.yocto.io/
>
> Maybe I should ask Michael or Joshua for help
>
> 2. The fix is on kernel-cache, not directly on oe-core,
> So it is difficult to use autobuilder
>
>Maybe I should ask Bruce for help
>


If you send a change for the kernel configuration, I can merge it, but not
activate it (i.e. I won't send SRCREV bumps), so as long as the autobuilder
can build a branch, we can queue this part of the change for testing.

Bruce


>
> 3. Most importantly, how to test, when find a way,
> the ideally is to add the test case to oe-core's
> meta-selftest
>
>I should do some investigation on the storage test.
>
> So I think I should do 3 first, and then 1/2 later
>
> //Hongxu
>
> Ross
>
>
>


-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end"
-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [OE-core] [PATCH 1/3] systemd: fix formatting IDE storage does not trigger "change" uevents

[Hongxu Jia]

If you send a change for the kernel configuration, I can merge it, but not
activate it (i.e. I won't send SRCREV bumps), so as long as the 
autobuilder

can build a branch, we can queue this part of the change for testing.



Got it, thanks

//Hongxu


Bruce


3. Most importantly, how to test, when find a way,
    the ideally is to add the test case to oe-core's
    meta-selftest

   I should do some investigation on the storage test.

So I think I should do 3 first, and then 1/2 later

//Hongxu


Ross






--
"Thou shalt not follow the NULL pointer, for chaos and madness await 
thee at its end"



-- 
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core