Skip to contentSkip to conversation

nvme-pci: disable write zeros support on SSDs

DISCUSSION Christoph Hellwig 2 participants 14 messages Sep 22, 2022 – Oct 4, 2022 1 message not in the archive NVMe

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Christoph Hellwig

author maintainer
#1 lore raw
Quoted 3 lines · Rita Lin

Dear receivers,

We would like to apply to disable write zeros support on Phison SSDs.

Can you explain why these SSDs do claim write Zeroes but it doesn't actually work, and why this is not fixable in a firmware update?

#2 lore raw

Dear Christoph,

Thank you for your reply.
These SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Thursday, September 22, 2022 1:32 PM
To: Rita Lin <[email protected]>
Cc: Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Christoph Hellwig <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Quoted 3 lines · Rita Lin

Dear receivers,

We would like to apply to disable write zeros support on Phison SSDs.

Can you explain why these SSDs do claim write Zeroes but it doesn't actually work, and why this is not fixable in a firmware update?

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

author maintainer
#3 lore raw
Quoted 6 lines · Rita Lin

Dear Christoph,

Thank you for your reply.
These SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

please add a comment explaining this to the IDs entry.

#4 lore raw

Dear Christoph,

Thank you for your advice.
Please find the updated patch in the attached file.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Thursday, September 22, 2022 1:53 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Quoted 6 lines · Rita Lin

Dear Christoph,

Thank you for your reply.
These SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

please add a comment explaining this to the IDs entry.

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

From 25a32a85ffa26cbbf0649d2b5457f6e13ec85903 Mon Sep 17 00:00:00 2001
From: Tina Hsu <[email protected]>
Date: Thu, 22 Sep 2022 14:16:30 +0800
Subject: [PATCH] [PATCH] Add Phison E3C/E4C Disable Write Zero

E3C/E4C SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

---
 drivers/nvme/host/pci.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/nvme/host/pci.c +4 -0
index 98864b853eef..67d3335e9cc8 100644
@@ -3470,6 +3470,10 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_DEVICE(0x1987, 0x5016),	/* Phison E16 */
 		.driver_data = NVME_QUIRK_IGNORE_DEV_SUBNQN |
 				NVME_QUIRK_BOGUS_NID, },
+	{ PCI_DEVICE(0x1987, 0x5019),  /* phison E19 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
+	{ PCI_DEVICE(0x1987, 0x5021),   /* Phison E21 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
 	{ PCI_DEVICE(0x1b4b, 0x1092),	/* Lexar 256 GB SSD */
 		.driver_data = NVME_QUIRK_NO_NS_DESC_LIST |
 				NVME_QUIRK_IGNORE_DEV_SUBNQN, },
Signature · 1 line
2.25.1
author maintainer
#5 lore raw

Can you give me a Signed-off-by: for this patch as well?

Also should this really use the phison.corp-partner.google.com or phison.com for the Author identification? (Same for the signoff).

#6 lore raw

Dear Christoph,

We had included signed off by information in the patch. We will use phison.corp-partner.google.com for this case. Thank you.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Thursday, September 22, 2022 10:26 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Can you give me a Signed-off-by: for this patch as well?

Also should this really use the phison.corp-partner.google.com or phison.com for the Author identification? (Same for the signoff).

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

From 25a32a85ffa26cbbf0649d2b5457f6e13ec85903 Mon Sep 17 00:00:00 2001
From: Tina Hsu <[email protected]>
Date: Thu, 22 Sep 2022 14:16:30 +0800
Subject: [PATCH] [PATCH] Add Phison E3C/E4C Disable Write Zero

E3C/E4C SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

---
 drivers/nvme/host/pci.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/nvme/host/pci.c +4 -0
index 98864b853eef..67d3335e9cc8 100644
@@ -3470,6 +3470,10 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_DEVICE(0x1987, 0x5016),	/* Phison E16 */
 		.driver_data = NVME_QUIRK_IGNORE_DEV_SUBNQN |
 				NVME_QUIRK_BOGUS_NID, },
+	{ PCI_DEVICE(0x1987, 0x5019),  /* phison E19 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
+	{ PCI_DEVICE(0x1987, 0x5021),   /* Phison E21 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
 	{ PCI_DEVICE(0x1b4b, 0x1092),	/* Lexar 256 GB SSD */
 		.driver_data = NVME_QUIRK_NO_NS_DESC_LIST |
 				NVME_QUIRK_IGNORE_DEV_SUBNQN, },
Signature · 1 line
2.25.1
author maintainer
#7 lore raw
Quoted 3 lines · Rita Lin

Dear Christoph,

We had included signed off by information in the patch.

I've not actually found a signoff in either version of the patch.

#8 lore raw

Dear Christoph,

The signed off information was put in line 10 in previous patch. I was not able to attach an image to identify the location to you.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Friday, September 23, 2022 2:12 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Quoted 3 lines · Rita Lin

Dear Christoph,

We had included signed off by information in the patch.

I've not actually found a signoff in either version of the patch.

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

author maintainer
#9 lore raw
Quoted 4 lines · Rita Lin

Dear Christoph,

The signed off information was put in line 10 in previous patch. I was not able to attach an image to identify the location to you.

I have not been able to find a Signed-off-by: line. Please provide one in reply to this mail.

#10 lore raw

Dear Christoph,

Please find the singed-off-by line below.

//signed-off-by: Tina Hsu <[email protected]>//

Reattach the patch again
Thanks

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Tuesday, September 27, 2022 3:08 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Quoted 4 lines · Rita Lin

Dear Christoph,

The signed off information was put in line 10 in previous patch. I was not able to attach an image to identify the location to you.

I have not been able to find a Signed-off-by: line. Please provide one in reply to this mail.

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

From 25a32a85ffa26cbbf0649d2b5457f6e13ec85903 Mon Sep 17 00:00:00 2001
From: Tina Hsu <[email protected]>
Date: Thu, 22 Sep 2022 14:16:30 +0800
Subject: [PATCH] [PATCH] Add Phison E3C/E4C Disable Write Zero

E3C/E4C SSDs do support write 0 command and the function was tested without issue.
But since these SSDs are DRAMless products, the write 0 command will take a lot longer to complete which had impacted the product performance. Due to the FW had been frozen, we would like to check if we could apply CL for these 2 products.

---
 drivers/nvme/host/pci.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/nvme/host/pci.c +4 -0
index 98864b853eef..67d3335e9cc8 100644
@@ -3470,6 +3470,10 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_DEVICE(0x1987, 0x5016),	/* Phison E16 */
 		.driver_data = NVME_QUIRK_IGNORE_DEV_SUBNQN |
 				NVME_QUIRK_BOGUS_NID, },
+	{ PCI_DEVICE(0x1987, 0x5019),  /* phison E19 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
+	{ PCI_DEVICE(0x1987, 0x5021),   /* Phison E21 */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
 	{ PCI_DEVICE(0x1b4b, 0x1092),	/* Lexar 256 GB SSD */
 		.driver_data = NVME_QUIRK_NO_NS_DESC_LIST |
 				NVME_QUIRK_IGNORE_DEV_SUBNQN, },
Signature · 1 line
2.25.1
author maintainer
#11 lore raw

Thanks, I've update the commit message a bit and applied it to the nvme-6.0 branch.

#12 lore raw

Hi Christoph

Would you please share the patch modified link on Gerrit Server to us?
In the meantime, would you please confirm if nvme-6.0 branch can be cherry-pick to all chrome branch? Thank you.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Tuesday, September 27, 2022 3:22 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Thanks, I've update the commit message a bit and applied it to the nvme-6.0 branch.

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.

author maintainer
#13 lore raw
Quoted 3 lines · Rita Lin

Hi Christoph

Would you please share the patch modified link on Gerrit Server to us?

I have no idea WTF you're even talking about.

#14 lore raw

This was the previous link we received as an example from others.
https://chromium-review.googlesource.com/c/chromiumos/third_party/kernel/+/3528180
Would like to understand if our application has something similar.

Rita Lin林萁煊
+886-37-586-896 Ext.80051
[email protected]

-----Original Message-----
From: Christoph Hellwig [mailto:[email protected]]
Sent: Tuesday, October 4, 2022 3:28 PM
To: Rita Lin <[email protected]>
Cc: Christoph Hellwig <[email protected]>; Keith Busch <[email protected]>; Jens Axboe <[email protected]>; Sagi Grimberg <[email protected]>; [email protected]; [email protected]; 黃晟瑋 <[email protected]>; Tina Hsu <[email protected]>; Vita Hsueh (薛伊倩) <[email protected]>; Wang, Audrey <[email protected]>
Subject: Re: nvme-pci: disable write zeros support on SSDs

Quoted 3 lines · Rita Lin

Hi Christoph

Would you please share the patch modified link on Gerrit Server to us?

I have no idea WTF you're even talking about.

This message and any attachments are confidential and may be legally privileged. Any unauthorized review, use or distribution by anyone other than the intended recipient is strictly prohibited. If you are not the intended recipient, please immediately notify the sender, completely delete the message and any attachments, and destroy all copies. Your cooperation will be highly appreciated.