Skip to contentSkip to conversation

e⋅MMC forward compatibility

PATCH Superseded Romain Izard 3 participants 5 messages Nov 28, 2013 – Jan 28, 2014 MMC, SD and SDIO

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

Hello,

The e⋅MMC 5.0 specification has been published by JEDEC on September 2013 [1], and contains a note in section B.8, which says: "A host system should work properly considering future eMMC version. For example, a host system is expected not to exit only due to the EXT_CSD_REV[192] value greater than 7 which will be used for next e•MMC revision."

Right now, the kernel controls the value of both fields 192 and 194, and rejects any device where CSD_STRUCTURE > 2 or EXT_CSD_REV > 7.

During the updates to the specification, JEDEC has maintained forward compatibility, setting reserved bits in the EXT_CSD register to describe new mandatory and optional features. If the specification changes in an incompatible way, it is expected that the value in CSD_STRUCTURE will change. As a result, I believe it would not be an issue to remove the control on EXT_CSD_REV.

I propose this change because e⋅MMC manufacturers are rapidly changing available references, and products with a long production life can encounter the situation where the kernel image, frozen at the beginning of the production, cannot support newer chips only because of this rule.

[1] http://www.jedec.org/standards-documents/docs/jesd84-b50 (Free registration required)

Best regards,

Signature · 1 line
Romain Izard
#1 lore raw

Hi Romain,

Quoted 9 lines · Romain Izard

The e⋅MMC 5.0 specification has been published by JEDEC on September 2013 [1], and contains a note in section B.8, which says: "A host system should work properly considering future eMMC version. For example, a host system is expected not to exit only due to the EXT_CSD_REV[192] value greater than 7 which will be used for next e•MMC revision."

Right now, the kernel controls the value of both fields 192 and 194, and rejects any device where CSD_STRUCTURE > 2 or EXT_CSD_REV > 7.

Sounds good to me -- want to send a patch?

Thanks,

- Chris.

Signature · 1 line
Chris Ball   <[email protected]>   <http://printf.net/>
#2 lore raw

PATCH 1/1 mmc: Allow forward compatibility for e⋅MMC

As stated by the e⋅MMC 5.0 specification, a chip should not be rejected only because of the revision stated in the EXT_CSD_REV field of the EXT_CSD register.

Remove the control on this value, the control of the CSD_STRUCTURE field should be sufficient to reject future incompatible changes.

---
 drivers/mmc/core/mmc.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/mmc/core/mmc.c +5 -6
index f631f5a9bf79..73b43574e1b4 100644
@@ -293,13 +293,12 @@ static int mmc_read_ext_csd(struct mmc_card *card, u8 *ext_csd)
 		}
 	}
 
+	/*
+	 * The EXT_CSD format is meant to be forward compatible. As long
+	 * as CSD_STRUCTURE does not change, all values for EXT_CSD_REV
+	 * are authorized, see JEDEC JESD84-B50 section B.8.
+	 */
 	card->ext_csd.rev = ext_csd[EXT_CSD_REV];
-	if (card->ext_csd.rev > 7) {
-		pr_err("%s: unrecognised EXT_CSD revision %d\n",
-			mmc_hostname(card->host), card->ext_csd.rev);
-		err = -EINVAL;
-		goto out;
-	}
 
 	card->ext_csd.raw_sectors[0] = ext_csd[EXT_CSD_SEC_CNT + 0];
 	card->ext_csd.raw_sectors[1] = ext_csd[EXT_CSD_SEC_CNT + 1];
Signature · 1 line
1.8.3.2
Acked-by on v1 1/1
#3 lore raw

Re: [PATCH] mmc: Allow forward compatibility for e⋅MMC

Quoted 8 lines · Romain Izard

As stated by the e⋅MMC 5.0 specification, a chip should not be rejected only because of the revision stated in the EXT_CSD_REV field of the EXT_CSD register.

Remove the control on this value, the control of the CSD_STRUCTURE field should be sufficient to reject future incompatible changes.

Quoted 30 lines
---
 drivers/mmc/core/mmc.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/mmc/core/mmc.c +5 -7
index f631f5a9bf79..73b43574e1b4 100644
@@ -293,13 +293,12 @@ static int mmc_read_ext_csd(struct mmc_card *card, u8 *ext_csd)
                }
        }
 
+       /*
+        * The EXT_CSD format is meant to be forward compatible. As long
+        * as CSD_STRUCTURE does not change, all values for EXT_CSD_REV
+        * are authorized, see JEDEC JESD84-B50 section B.8.
+        */
        card->ext_csd.rev = ext_csd[EXT_CSD_REV];
-       if (card->ext_csd.rev > 7) {
-               pr_err("%s: unrecognised EXT_CSD revision %d\n",
-                       mmc_hostname(card->host), card->ext_csd.rev);
-               err = -EINVAL;
-               goto out;
-       }
 
        card->ext_csd.raw_sectors[0] = ext_csd[EXT_CSD_SEC_CNT + 0];
        card->ext_csd.raw_sectors[1] = ext_csd[EXT_CSD_SEC_CNT + 1];
--

1.8.3.2

#4 lore raw

PATCH RESEND 1/1 mmc: Allow forward compatibility for e⋅MMC

In reply to Romain Izard

As stated by the e⋅MMC 5.0 specification, a chip should not be rejected only because of the revision stated in the EXT_CSD_REV field of the EXT_CSD register.

Remove the control on this value, the control of the CSD_STRUCTURE field should be sufficient to reject future incompatible changes.

---
 drivers/mmc/core/mmc.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/mmc/core/mmc.c +5 -6
index 98e9eb0f6643..81c23e4db704 100644
@@ -293,13 +293,12 @@ static int mmc_read_ext_csd(struct mmc_card *card, u8 *ext_csd)
 		}
 	}
 
+	/*
+	 * The EXT_CSD format is meant to be forward compatible. As long
+	 * as CSD_STRUCTURE does not change, all values for EXT_CSD_REV
+	 * are authorized, see JEDEC JESD84-B50 section B.8.
+	 */
 	card->ext_csd.rev = ext_csd[EXT_CSD_REV];
-	if (card->ext_csd.rev > 7) {
-		pr_err("%s: unrecognised EXT_CSD revision %d\n",
-			mmc_hostname(card->host), card->ext_csd.rev);
-		err = -EINVAL;
-		goto out;
-	}
 
 	card->ext_csd.raw_sectors[0] = ext_csd[EXT_CSD_SEC_CNT + 0];
 	card->ext_csd.raw_sectors[1] = ext_csd[EXT_CSD_SEC_CNT + 1];
Signature · 1 line
1.8.3.2