mail archive of the barebox mailing list
 help / color / mirror / Atom feed
From: John Watts <contact@jookia.org>
To: barebox@lists.infradead.org
Cc: John Watts <contact@jookia.org>
Subject: [PATCH v2 1/2] imx6-mmdc: Work around ERR050070
Date: Wed, 25 Jan 2023 04:41:15 +1100	[thread overview]
Message-ID: <20230124174115.1509216-1-contact@jookia.org> (raw)
In-Reply-To: <20230123201817.1084728-1-contact@jookia.org>

The MPWLGCR registers clear the error bits before software can read them,
so rely on the MPWLHWERR registers for error reporting instead.

This errata was announced in 2019 but it seems to apply to all chip revisions.

U-Boot contains a different workaround where it instead checks the calibration
data for the result 0x001F001F (maximum delay) and flag that as a failure.
I can't find the origin of this workaround but I first saw it in the Novena
source code, though I asked Sean Cross and he suggested it was from Freescale.

While we're at it, fix the comment implying this code only checks PHY0 in x32
configuration. This is wrong and misleading.

Signed-off-by: John Watts <contact@jookia.org>
---
Changes v1 -> v2:
- Added a wall of text explaining what the code does and why
- wlcalib_failed now reads from the ips register correctly
---
 arch/arm/mach-imx/imx6-mmdc.c              | 30 +++++++++++++++++++---
 arch/arm/mach-imx/include/mach/imx6-mmdc.h |  1 +
 2 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/arch/arm/mach-imx/imx6-mmdc.c b/arch/arm/mach-imx/imx6-mmdc.c
index 00b8d30d69..45e1b030d3 100644
--- a/arch/arm/mach-imx/imx6-mmdc.c
+++ b/arch/arm/mach-imx/imx6-mmdc.c
@@ -11,6 +11,30 @@
 #include <mach/imx6-regs.h>
 #include <mach/imx6.h>
 
+static bool wlcalib_failed(void __iomem *ips)
+{
+	/*
+	 * The i.MX 6 reference manual specifies that an MMDC flags reports
+	 * write calibration errors in the MPWLGCR register's HW_WL_ERR field.
+	 *
+	 * ERR050070 specifies that this doesn't work and we should check
+	 * the MPWLHWERR register instead which reports which write leveling
+	 * steps succeeded or failed on a per-byte basis.
+	 *
+	 * Check each byte to see which steps succeeded. If no steps succeeded
+	 * then declare the calibration a failure.
+	*/
+
+	int i;
+
+	for (i = 0; i < 4; ++i) {
+		if (readb(ips + MPWLHWERR + i) == 0)
+			return true;
+	}
+
+	return false;
+}
+
 int mmdc_do_write_level_calibration(void)
 {
 	u32 esdmisc_val, zq_val;
@@ -56,11 +80,9 @@ int mmdc_do_write_level_calibration(void)
 	/* Upon completion of this process the MMDC de-asserts the MPWLGCR[HW_WL_EN] */
 	while (readl(P0_IPS + MPWLGCR) & 0x00000001);
 
-	/* check for any errors: check both PHYs for x64 configuration, if x32, check only PHY0 */
-	if ((readl(P0_IPS + MPWLGCR) & 0x00000F00) ||
-			(readl(P1_IPS + MPWLGCR) & 0x00000F00)) {
+	/* check for any errors on both PHYs */
+	if (wlcalib_failed(P0_IPS) || wlcalib_failed(P1_IPS))
 		errorcount++;
-	}
 
 	pr_debug("Write leveling calibration completed\n");
 
diff --git a/arch/arm/mach-imx/include/mach/imx6-mmdc.h b/arch/arm/mach-imx/include/mach/imx6-mmdc.h
index bda20aba17..098ba4f5bf 100644
--- a/arch/arm/mach-imx/include/mach/imx6-mmdc.h
+++ b/arch/arm/mach-imx/include/mach/imx6-mmdc.h
@@ -18,6 +18,7 @@
 #define MPWLGCR		0x808
 #define MPWLDECTRL0	0x80c
 #define MPWLDECTRL1	0x810
+#define MPWLHWERR	0x878
 #define MPPDCMPR1	0x88c
 #define MPSWDAR		0x894
 #define MPRDDLCTL	0x848
-- 
2.39.0




  parent reply	other threads:[~2023-01-24 17:44 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-23 20:18 [PATCH " John Watts
2023-01-23 20:18 ` [PATCH 2/2] imx6-mmdc: Revert calibration configuration on failure John Watts
2023-01-24  8:29   ` Ahmad Fatoum
2023-01-24 12:01     ` John Watts
2023-01-24 12:07       ` Ahmad Fatoum
2023-01-24  8:27 ` [PATCH 1/2] imx6-mmdc: Work around ERR050070 Ahmad Fatoum
2023-01-24 11:48   ` John Watts
2023-01-24 11:58     ` Ahmad Fatoum
2023-01-24 17:41 ` John Watts [this message]
2023-01-25 15:48   ` [PATCH v2 " Sascha Hauer
2023-01-24 17:41 ` [PATCH v2 2/2] imx6-mmdc: Revert calibration configuration on failure John Watts

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230124174115.1509216-1-contact@jookia.org \
    --to=contact@jookia.org \
    --cc=barebox@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox