Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except one issue: the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
This is what I did:
U-Boot> setenv pio_kernel "recovery/kernel.img" U-Boot> setenv pio_initrd "recovery/initrd.img" U-Boot> fatload mmc 0:1 ${kernel_addr_r} ${pio_kernel} 5299624 bytes read in 45262 ms (114.3 KiB/s) U-Boot> fatload mmc 0:1 ${ramdisk_addr_r} ${pio_initrd} 24239932 bytes read in 206978 ms (114.3 KiB/s)
The configuration of by build is attached.
Is there anyone else experiencing this kind of issues?
Thank you very much.
Mario
Mario,
-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Mario Lombardo Sent: Wednesday, November 3, 2021 5:40 AM To: u-boot@lists.denx.de Subject: [External] - Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except one issue: the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
Please see this series: https://patchwork.ozlabs.org/project/uboot/list/?series=262375 And this RPi issue: https://github.com/raspberrypi/firmware/issues/1619
This is what I did:
U-Boot> setenv pio_kernel "recovery/kernel.img" U-Boot> setenv pio_initrd "recovery/initrd.img" U-Boot> fatload mmc 0:1 ${kernel_addr_r} ${pio_kernel} 5299624 bytes read in 45262 ms (114.3 KiB/s) U-Boot> fatload mmc 0:1 ${ramdisk_addr_r} ${pio_initrd} 24239932 bytes read in 206978 ms (114.3 KiB/s)
The configuration of by build is attached.
Is there anyone else experiencing this kind of issues?
Thank you very much.
Mario
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
-Vincent
Vincent,
thank you very much. Applying the two patches mentioned in the series resolve the issues immediately.
Is it expected to have this enhancement in the upcoming u-boot release?
Cheers, Mario
On 3. Nov 2021, at 21:57, Vincent Fazio vfazio@xes-inc.com wrote:
Mario,
-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Mario Lombardo Sent: Wednesday, November 3, 2021 5:40 AM To: u-boot@lists.denx.de Subject: [External] - Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except one issue: the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
Please see this series: https://patchwork.ozlabs.org/project/uboot/list/?series=262375 And this RPi issue: https://github.com/raspberrypi/firmware/issues/1619
This is what I did:
U-Boot> setenv pio_kernel "recovery/kernel.img" U-Boot> setenv pio_initrd "recovery/initrd.img" U-Boot> fatload mmc 0:1 ${kernel_addr_r} ${pio_kernel} 5299624 bytes read in 45262 ms (114.3 KiB/s) U-Boot> fatload mmc 0:1 ${ramdisk_addr_r} ${pio_initrd} 24239932 bytes read in 206978 ms (114.3 KiB/s)
The configuration of by build is attached.
Is there anyone else experiencing this kind of issues?
Thank you very much.
Mario
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
-Vincent
——— Mario Lombardo akamo GmbH | Eberhardstraße 65 | 70173 Stuttgart @: ml@akamo.de t: +49-711-25295970 w: www.akamo.de
Amtsgericht Stuttgart, HRB 769814 Geschäftsführer: Martin Filz, Mario Lombardo, Tobias Ott
Diese E-Mail enthält vertrauliche und / oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese E-Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser E-Mail ist nicht gestattet.
On 11-03 20:57, Vincent Fazio wrote:
Date: Wed, 3 Nov 2021 20:57:13 +0000 From: Vincent Fazio vfazio@xes-inc.com To: Mario Lombardo ml@akamo.de, "u-boot@lists.denx.de" u-boot@lists.denx.de Subject: RE: [External] - Slow MMC IO with Raspberry CM3+ Module Message-ID: BN1P110MB09373DD94630D64646C46DF69F8C9@BN1P110MB0937.NAMP110.PROD.OUTLOOK.COM
Tags: all uboot
Hi Vincent,
Mario,
-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Mario Lombardo Sent: Wednesday, November 3, 2021 5:40 AM To: u-boot@lists.denx.de Subject: [External] - Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except one issue: the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
Please see this series: https://patchwork.ozlabs.org/project/uboot/list/?series=262375 And this RPi issue: https://github.com/raspberrypi/firmware/issues/1619
Are your patches still needed? If I read the comments on the issue it was fixed in RPi firmware and your workaround is not needed, correct?
Regards, Ivan
Ivan,
-----Original Message----- From: Ivan T. Ivanov iivanov@suse.de Sent: Wednesday, November 10, 2021 4:01 AM To: Vincent Fazio vfazio@xes-inc.com Cc: Mario Lombardo ml@akamo.de; u-boot@lists.denx.de Subject: Re: [External] - Slow MMC IO with Raspberry CM3+ Module
On 11-03 20:57, Vincent Fazio wrote:
Date: Wed, 3 Nov 2021 20:57:13 +0000 From: Vincent Fazio vfazio@xes-inc.com To: Mario Lombardo ml@akamo.de, "u-boot@lists.denx.de" u-boot@lists.denx.de Subject: RE: [External] - Slow MMC IO with Raspberry CM3+ Module Message-ID:
<BN1P110MB09373DD94630D64646C46DF69F8C9@BN1P110MB0937.NAMP11 0.PROD.OUT
LOOK.COM>
Tags: all uboot
Hi Vincent,
Mario,
-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Mario Lombardo Sent: Wednesday, November 3, 2021 5:40 AM To: u-boot@lists.denx.de Subject: [External] - Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except
one issue:
the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
Please see this series: https://usg02.safelinks.protection.office365.us/?url=https%3A%2F%2Fpat
chwork.ozlabs.org%2Fproject%2Fuboot%2Flist%2F%3Fseries%3D262375&am p;da
ta=04%7C01%7C%7Ca736ee7a8d7446c34adb08d9a430fdf3%7C2925f1cdbdc34 a76bb3
86159e20a17f1%7C0%7C0%7C637721352603649117%7CUnknown%7CTWFpb GZsb3d8eyJ
WIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D% 7C1000
&sdata=4snQV571x4aGlf7NwXnR%2B%2F5RgkyZ2MOMax%2FudbP6xu w%3D&re
served=0 And this RPi issue: https://usg02.safelinks.protection.office365.us/?url=https%3A%2F%2Fgit
hub.com%2Fraspberrypi%2Ffirmware%2Fissues%2F1619&data=04%7C0 1%7C%7
Ca736ee7a8d7446c34adb08d9a430fdf3%7C2925f1cdbdc34a76bb386159e20a1 7f1%7
C0%7C0%7C637721352603659112%7CUnknown%7CTWFpbGZsb3d8eyJWIjoi MC4wLjAwMD
AiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata= das1
lOBGq%2BMDOz7OlqbEbyBS9eXtrp7W5SQ25VteYLk%3D&reserved=0
Are your patches still needed? If I read the comments on the issue it was fixed in RPi firmware and your workaround is not needed, correct?
I think this is a fair question.
The patches are still needed for certain tagged versions of the RPi firmware. Later tags may resolve this problem but may also introduce their own regressions which force users back to a previous tag where this is a problem.
I think I would recommend including the patches so U-Boot works across all revisions of firmware and to future proof against subsequent changes in the clock API.
Ultimately, I defer to the maintainers' choice.
Regards, Ivan
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Vincent Fazio Embedded Software Engineer - ATS Extreme Engineering Solutions, Inc http://www.xes-inc.com
Hi Vincent
On 11/11/21 1:45 AM, Vincent Fazio wrote:
Ivan,
-----Original Message----- From: Ivan T. Ivanov iivanov@suse.de Sent: Wednesday, November 10, 2021 4:01 AM To: Vincent Fazio vfazio@xes-inc.com Cc: Mario Lombardo ml@akamo.de; u-boot@lists.denx.de Subject: Re: [External] - Slow MMC IO with Raspberry CM3+ Module
On 11-03 20:57, Vincent Fazio wrote:
Date: Wed, 3 Nov 2021 20:57:13 +0000 From: Vincent Fazio vfazio@xes-inc.com To: Mario Lombardo ml@akamo.de, "u-boot@lists.denx.de" u-boot@lists.denx.de Subject: RE: [External] - Slow MMC IO with Raspberry CM3+ Module Message-ID:
<BN1P110MB09373DD94630D64646C46DF69F8C9@BN1P110MB0937.NAMP11 0.PROD.OUT
LOOK.COM>
Tags: all uboot
Hi Vincent,
Mario,
-----Original Message----- From: U-Boot u-boot-bounces@lists.denx.de On Behalf Of Mario Lombardo Sent: Wednesday, November 3, 2021 5:40 AM To: u-boot@lists.denx.de Subject: [External] - Slow MMC IO with Raspberry CM3+ Module
Dear u-boot list,
I recently compiled (latest: 2021.10) code of uboot as a boot loader for the Compute Module 3+ (Raspberry). Everything works fine except
one issue:
the IO from the storage is extreme slow. I read about issues in the past related to ext file systems. However the present issue relates to both ext and fat file systems so I assume its cause is different.
Please see this series: https://usg02.safelinks.protection.office365.us/?url=https%3A%2F%2Fpat
chwork.ozlabs.org%2Fproject%2Fuboot%2Flist%2F%3Fseries%3D262375&am p;da
ta=04%7C01%7C%7Ca736ee7a8d7446c34adb08d9a430fdf3%7C2925f1cdbdc34 a76bb3
86159e20a17f1%7C0%7C0%7C637721352603649117%7CUnknown%7CTWFpb GZsb3d8eyJ
WIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D% 7C1000
&sdata=4snQV571x4aGlf7NwXnR%2B%2F5RgkyZ2MOMax%2FudbP6xu w%3D&re
served=0 And this RPi issue: https://usg02.safelinks.protection.office365.us/?url=https%3A%2F%2Fgit
hub.com%2Fraspberrypi%2Ffirmware%2Fissues%2F1619&data=04%7C0 1%7C%7
Ca736ee7a8d7446c34adb08d9a430fdf3%7C2925f1cdbdc34a76bb386159e20a1 7f1%7
C0%7C0%7C637721352603659112%7CUnknown%7CTWFpbGZsb3d8eyJWIjoi MC4wLjAwMD
AiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata= das1
lOBGq%2BMDOz7OlqbEbyBS9eXtrp7W5SQ25VteYLk%3D&reserved=0
Are your patches still needed? If I read the comments on the issue it was fixed in RPi firmware and your workaround is not needed, correct?
I think this is a fair question.
The patches are still needed for certain tagged versions of the RPi firmware. Later tags may resolve this problem but may also introduce their own regressions which force users back to a previous tag where this is a problem.
I think I would recommend including the patches so U-Boot works across all revisions of firmware and to future proof against subsequent changes in the clock API.
Ultimately, I defer to the maintainers' choice.
Sorry, I missed your patch. I will test with your patch on today. I think that it's good to apply yours if there is no side-effect or any problem.
Best Regards, Jaehoon Chung
Regards, Ivan
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Vincent Fazio Embedded Software Engineer - ATS Extreme Engineering Solutions, Inc https://protect2.fireeye.com/v1/url?k=71bc6382-2e275aad-71bde8cd-0cc47a31cdf...
participants (4)
-
Ivan T. Ivanov -
Jaehoon Chung -
Mario Lombardo -
Vincent Fazio