All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v4 00/16] Add support for NVMEM API
@ 2022-05-05 17:11 Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices Sean Anderson
                   ` (15 more replies)
  0 siblings, 16 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson, Alex Marginean, Claudiu Manoil,
	Marek Behún, Vladimir Oltean

This adds support for the nvmem-cells properties cropping up in manyb
device trees. This is an easy way to load configuration, version
information, or calibration data from a non-volatile memory source. For
more information, refer to patch 6 ("misc: Add support for nvmem
cells").

For the moment I have only added some integration tests using the
ethernet addresses. This hits the main code paths (looking up nvmem
cells) but doesn't test writing. I can add a few stand-alone tests if
desired.

CI: https://dev.azure.com/u-boot/u-boot/_build/results?buildId=4167&view=results

The patches are structured in the following manner:
1-9:   These are general cleanups, and may be applied independently of
       the rest of the series.
10-12: Add NVMEM support
13-15: Support reading ethernet addresses using the NVMEM API and add
       some tests.
16:    Example demonstrating how this series could be reworked to be a
       uclass. It should not be applied.

Changes in v4:
- Fix dm_test_alias_highest_id test failing because we changed the
  number of aliases
- Fix segmentation fault if DSA master fails to probe
- Add fallbacks for i2c_eeprom_* functions. This fixes build errors if
  CONFIG_I2C_EEPROM was not enabled.
- Fix failing to find nvmem devices if some supported uclasses were not
  enabled.
- Remove typedefs from documentation
- Document some of the design background, and why a new uclass was not
  created.
- Enable CONFIG_NVMEM for sandbox_spl_defconfig
- Enable CONFIG_I2C_EEPROM for sandbox_spl_defconfig
- Add RFC patch illustrating how this series could be reworked to use a
  more conventional, uclass-based approach.

Changes in v3:
- Add aliases for ethernet devices
- Add mac address for eth8 to environment
- Move patch adding test earlier in the series
- Add test for eth8 as well

Changes in v2:
- Call the appropriate API functions directly from
  nvmem_cell_(read|write). This means we can drop the nvmem_interface
  machinery.

Sean Anderson (16):
  sandbox: net: Add aliases for ethernet devices
  sandbox: net: Add mac address for eth8 to environment
  test: eth: Add test for ethernet addresses
  sandbox: net: Remove fake-host-hwaddr
  sandbox: Remove eth2addr from environment
  sandbox: Move some mac addresses to device tree
  net: dsa: Fix segmentation fault if master fails to probe
  misc: i2c_eeprom: Make i2c_eeprom_write use a const buf
  misc: i2c_eeprom: Add fallbacks
  misc: Add support for nvmem cells
  sandbox: Enable NVMEM
  net: Add support for reading mac addresses from nvmem cells
  test: Load mac address with i2c eeprom
  test: Load mac address using RTC
  test: Load mac address using misc device
  [RFC] misc: nvmem: Convert to using udevices

 MAINTAINERS                        |   7 ++
 arch/sandbox/dts/sandbox.dts       |   1 -
 arch/sandbox/dts/sandbox64.dts     |   1 -
 arch/sandbox/dts/test.dts          |  34 +++++-
 board/sandbox/sandbox.env          |   4 -
 configs/sandbox64_defconfig        |   1 +
 configs/sandbox_defconfig          |   1 +
 configs/sandbox_flattree_defconfig |   1 +
 configs/sandbox_noinst_defconfig   |   1 +
 configs/sandbox_spl_defconfig      |   3 +
 doc/api/index.rst                  |   1 +
 doc/api/nvmem.rst                  |   7 ++
 drivers/misc/Kconfig               |  16 +++
 drivers/misc/Makefile              |   1 +
 drivers/misc/i2c_eeprom.c          |  40 ++++++-
 drivers/misc/i2c_eeprom_emul.c     |   4 +
 drivers/misc/misc-uclass.c         |  58 ++++++++-
 drivers/misc/misc_sandbox.c        |   3 +
 drivers/misc/nvmem.c               | 106 +++++++++++++++++
 drivers/net/sandbox.c              |  10 +-
 drivers/rtc/i2c_rtc_emul.c         |  10 ++
 drivers/rtc/rtc-uclass.c           |  46 ++++++-
 include/dm/uclass-id.h             |   1 +
 include/i2c_eeprom.h               |  27 ++++-
 include/nvmem.h                    | 185 +++++++++++++++++++++++++++++
 net/dsa-uclass.c                   |   6 +-
 net/eth-uclass.c                   |  13 +-
 test/dm/eth.c                      |  29 +++++
 test/dm/test-fdt.c                 |   2 +-
 29 files changed, 584 insertions(+), 35 deletions(-)
 create mode 100644 doc/api/nvmem.rst
 create mode 100644 drivers/misc/nvmem.c
 create mode 100644 include/nvmem.h

-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply	[flat|nested] 19+ messages in thread

* [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-06-09 19:21   ` Tom Rini
  2022-05-05 17:11 ` [PATCH v4 02/16] sandbox: net: Add mac address for eth8 to environment Sean Anderson
                   ` (14 subsequent siblings)
  15 siblings, 1 reply; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson, Marek Behún

Commit f3dd213e15 ("net: introduce helpers to get PHY ofnode from MAC")
changed the ethernet sequence assignment from

uclass 36: ethernet
0   * eth@10002000 @ 05813460, seq 0
1   * eth@10003000 @ 05813550, seq 5
2   * sbe5 @ 05813640, seq 3
3   * eth@10004000 @ 05813730, seq 6
4   * dsa-test-eth @ 05813820, seq 4
5   * lan0 @ 05813a30, seq 2
6   * lan1 @ 05813b50, seq 7

to

uclass 36: ethernet
0   * eth@10002000 @ 03813630, seq 0
1   * eth@10003000 @ 03813720, seq 5
2   * sbe5 @ 03813810, seq 3
3   * eth@10004000 @ 03813900, seq 6
4     phy-test-eth @ 038139f0, seq 7
5   * dsa-test-eth @ 03813ae0, seq 4
6   * lan0 @ 03813cf0, seq 2
7   * lan1 @ 03813e10, seq 8

This caused the mac address assignment to switch around. Avoid this in
the future by assigning aliases for all ethernet devices. This reverts
the sequence to what it was before the aformentioned commit (with
phy-test-eth as seq 8). There is no ethernet1 for whatever reason.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

Changes in v4:
- Fix dm_test_alias_highest_id test failing because we changed the
  number of aliases

Changes in v3:
- New

 arch/sandbox/dts/test.dts | 3 +++
 test/dm/test-fdt.c        | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index a8a86bc715..c0b1d76cdb 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -28,6 +28,9 @@
 		ethernet3 = &eth_3;
 		ethernet4 = &dsa_eth0;
 		ethernet5 = &eth_5;
+		ethernet6 = "/eth@10004000";
+		ethernet7 = &swp_1;
+		ethernet8 = &phy_eth0;
 		gpio1 = &gpio_a;
 		gpio2 = &gpio_b;
 		gpio3 = &gpio_c;
diff --git a/test/dm/test-fdt.c b/test/dm/test-fdt.c
index e1de066226..f9e8174759 100644
--- a/test/dm/test-fdt.c
+++ b/test/dm/test-fdt.c
@@ -184,7 +184,7 @@ static int dm_test_alias_highest_id(struct unit_test_state *uts)
 	int ret;
 
 	ret = dev_read_alias_highest_id("ethernet");
-	ut_asserteq(5, ret);
+	ut_asserteq(8, ret);
 
 	ret = dev_read_alias_highest_id("gpio");
 	ut_asserteq(3, ret);
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 02/16] sandbox: net: Add mac address for eth8 to environment
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 03/16] test: eth: Add test for ethernet addresses Sean Anderson
                   ` (13 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson, Marek Behún

The phy_eth0 interface introduced in commit f3dd213e15 ("net: introduce
helpers to get PHY ofnode from MAC") uses a globally-administered
address. Switch to using a locally-administered address, and add it to
the sandbox environment, like the others.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

Changes in v4:
- Add "commit" in front of the hash

Changes in v3:
- New

 board/sandbox/sandbox.env | 1 +
 1 file changed, 1 insertion(+)

diff --git a/board/sandbox/sandbox.env b/board/sandbox/sandbox.env
index b4c04635a4..6dedc755fa 100644
--- a/board/sandbox/sandbox.env
+++ b/board/sandbox/sandbox.env
@@ -11,6 +11,7 @@ eth3addr=02:00:11:22:33:45
 eth4addr=02:00:11:22:33:48
 eth5addr=02:00:11:22:33:46
 eth6addr=02:00:11:22:33:47
+eth8addr=02:00:11:22:33:49
 ipaddr=192.0.2.1
 
 /*
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 03/16] test: eth: Add test for ethernet addresses
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 02/16] sandbox: net: Add mac address for eth8 to environment Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 04/16] sandbox: net: Remove fake-host-hwaddr Sean Anderson
                   ` (12 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This adds a test to make sure that all the ethernet interfaces have
their addresses read properly. At the moment everything is read from the
environment, but the next few commits will add additional sources.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v3)

Changes in v3:
- Move patch adding test earlier in the series
- Add test for eth8 as well

 test/dm/eth.c | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

diff --git a/test/dm/eth.c b/test/dm/eth.c
index e4ee695610..5437f9ea4a 100644
--- a/test/dm/eth.c
+++ b/test/dm/eth.c
@@ -147,6 +147,35 @@ static int dm_test_eth_act(struct unit_test_state *uts)
 }
 DM_TEST(dm_test_eth_act, UT_TESTF_SCAN_FDT);
 
+/* Ensure that all addresses are loaded properly */
+static int dm_test_ethaddr(struct unit_test_state *uts)
+{
+	static const char *const addr[] = {
+		"02:00:11:22:33:44",
+		"02:00:11:22:33:48", /* dsa slave */
+		"02:00:11:22:33:45",
+		"02:00:11:22:33:48", /* dsa master */
+		"02:00:11:22:33:46",
+		"02:00:11:22:33:47",
+		"02:00:11:22:33:48", /* dsa slave */
+		"02:00:11:22:33:49",
+	};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(addr); i++) {
+		char addrname[10];
+
+		if (i)
+			snprintf(addrname, sizeof(addrname), "eth%daddr", i + 1);
+		else
+			strcpy(addrname, "ethaddr");
+		ut_asserteq_str(addr[i], env_get(addrname));
+	}
+
+	return 0;
+}
+DM_TEST(dm_test_ethaddr, UT_TESTF_SCAN_FDT);
+
 /* The asserts include a return on fail; cleanup in the caller */
 static int _dm_test_eth_rotate1(struct unit_test_state *uts)
 {
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 04/16] sandbox: net: Remove fake-host-hwaddr
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (2 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 03/16] test: eth: Add test for ethernet addresses Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 05/16] sandbox: Remove eth2addr from environment Sean Anderson
                   ` (11 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

Instead of reading a pseudo-rom mac address from the device tree, just use
whatever we get from write_hwaddr. This has the effect of using the mac
address from the environment (or from the device tree, if it is
specified).

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Ramon Fried <rfried.dev@gmail.com>
---

(no changes since v1)

 arch/sandbox/dts/sandbox.dts   |  1 -
 arch/sandbox/dts/sandbox64.dts |  1 -
 arch/sandbox/dts/test.dts      |  6 ------
 drivers/net/sandbox.c          | 10 ++--------
 4 files changed, 2 insertions(+), 16 deletions(-)

diff --git a/arch/sandbox/dts/sandbox.dts b/arch/sandbox/dts/sandbox.dts
index 18fde1c8c6..21f00fcab5 100644
--- a/arch/sandbox/dts/sandbox.dts
+++ b/arch/sandbox/dts/sandbox.dts
@@ -63,7 +63,6 @@
 	eth@10002000 {
 		compatible = "sandbox,eth";
 		reg = <0x10002000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 00];
 	};
 
 	host-fs {
diff --git a/arch/sandbox/dts/sandbox64.dts b/arch/sandbox/dts/sandbox64.dts
index ec53106af9..3eb0457089 100644
--- a/arch/sandbox/dts/sandbox64.dts
+++ b/arch/sandbox/dts/sandbox64.dts
@@ -58,7 +58,6 @@
 	eth@10002000 {
 		compatible = "sandbox,eth";
 		reg = <0x0 0x10002000 0x0 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 00];
 	};
 
 	i2c_0: i2c@0 {
diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index c0b1d76cdb..e486bdb531 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -527,31 +527,26 @@
 	eth@10002000 {
 		compatible = "sandbox,eth";
 		reg = <0x10002000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 00];
 	};
 
 	eth_5: eth@10003000 {
 		compatible = "sandbox,eth";
 		reg = <0x10003000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 11];
 	};
 
 	eth_3: sbe5 {
 		compatible = "sandbox,eth";
 		reg = <0x10005000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 33];
 	};
 
 	eth@10004000 {
 		compatible = "sandbox,eth";
 		reg = <0x10004000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 22];
 	};
 
 	phy_eth0: phy-test-eth {
 		compatible = "sandbox,eth";
 		reg = <0x10007000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 77];
 		phy-handle = <&ethphy1>;
 		phy-mode = "2500base-x";
 	};
@@ -559,7 +554,6 @@
 	dsa_eth0: dsa-test-eth {
 		compatible = "sandbox,eth";
 		reg = <0x10006000 0x1000>;
-		fake-host-hwaddr = [00 00 66 44 22 66];
 	};
 
 	dsa-test {
diff --git a/drivers/net/sandbox.c b/drivers/net/sandbox.c
index 37459dfa0a..13022addb6 100644
--- a/drivers/net/sandbox.c
+++ b/drivers/net/sandbox.c
@@ -395,9 +395,11 @@ static void sb_eth_stop(struct udevice *dev)
 static int sb_eth_write_hwaddr(struct udevice *dev)
 {
 	struct eth_pdata *pdata = dev_get_plat(dev);
+	struct eth_sandbox_priv *priv = dev_get_priv(dev);
 
 	debug("eth_sandbox %s: Write HW ADDR - %pM\n", dev->name,
 	      pdata->enetaddr);
+	memcpy(priv->fake_host_hwaddr, pdata->enetaddr, ARP_HLEN);
 	return 0;
 }
 
@@ -419,16 +421,8 @@ static int sb_eth_of_to_plat(struct udevice *dev)
 {
 	struct eth_pdata *pdata = dev_get_plat(dev);
 	struct eth_sandbox_priv *priv = dev_get_priv(dev);
-	const u8 *mac;
 
 	pdata->iobase = dev_read_addr(dev);
-
-	mac = dev_read_u8_array_ptr(dev, "fake-host-hwaddr", ARP_HLEN);
-	if (!mac) {
-		printf("'fake-host-hwaddr' is missing from the DT\n");
-		return -EINVAL;
-	}
-	memcpy(priv->fake_host_hwaddr, mac, ARP_HLEN);
 	priv->disabled = false;
 	priv->tx_handler = sb_default_handler;
 
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 05/16] sandbox: Remove eth2addr from environment
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (3 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 04/16] sandbox: net: Remove fake-host-hwaddr Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 06/16] sandbox: Move some mac addresses to device tree Sean Anderson
                   ` (10 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

DSA interfaces use the same mac address for each interface, unless
instructed otherwise. Just set eth4addr and let eth2addr and eth7addr be
set automatically.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 board/sandbox/sandbox.env | 1 -
 1 file changed, 1 deletion(-)

diff --git a/board/sandbox/sandbox.env b/board/sandbox/sandbox.env
index 6dedc755fa..88ed7a9606 100644
--- a/board/sandbox/sandbox.env
+++ b/board/sandbox/sandbox.env
@@ -6,7 +6,6 @@ stdout=serial,vidconsole
 stderr=serial,vidconsole
 
 ethaddr=02:00:11:22:33:44
-eth2addr=02:00:11:22:33:48
 eth3addr=02:00:11:22:33:45
 eth4addr=02:00:11:22:33:48
 eth5addr=02:00:11:22:33:46
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 06/16] sandbox: Move some mac addresses to device tree
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (4 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 05/16] sandbox: Remove eth2addr from environment Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe Sean Anderson
                   ` (9 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This prevents some conflicts when running sandbox with -D, since the
"rom" mac address will be random and won't match the environment. We
still need to keep addresses for eth1 and eth6 in the environment,
because dm_test_eth_rotate expects to be able to disable them by
removing their envaddr variables. This can likely be fixed in a future
series by adding a function to cause sandbox eth_opts callback for a
particular mac to fail immediately.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 arch/sandbox/dts/test.dts | 4 ++++
 board/sandbox/sandbox.env | 4 ----
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index e486bdb531..318dc2dcb1 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -532,11 +532,13 @@
 	eth_5: eth@10003000 {
 		compatible = "sandbox,eth";
 		reg = <0x10003000 0x1000>;
+		mac-address = [ 02 00 11 22 33 46 ];
 	};
 
 	eth_3: sbe5 {
 		compatible = "sandbox,eth";
 		reg = <0x10005000 0x1000>;
+		mac-address = [ 02 00 11 22 33 45 ];
 	};
 
 	eth@10004000 {
@@ -547,6 +549,7 @@
 	phy_eth0: phy-test-eth {
 		compatible = "sandbox,eth";
 		reg = <0x10007000 0x1000>;
+		mac-address = [ 02 00 11 22 33 49 ];
 		phy-handle = <&ethphy1>;
 		phy-mode = "2500base-x";
 	};
@@ -554,6 +557,7 @@
 	dsa_eth0: dsa-test-eth {
 		compatible = "sandbox,eth";
 		reg = <0x10006000 0x1000>;
+		mac-address = [ 02 00 11 22 33 48 ];
 	};
 
 	dsa-test {
diff --git a/board/sandbox/sandbox.env b/board/sandbox/sandbox.env
index 88ed7a9606..a2c19702d6 100644
--- a/board/sandbox/sandbox.env
+++ b/board/sandbox/sandbox.env
@@ -6,11 +6,7 @@ stdout=serial,vidconsole
 stderr=serial,vidconsole
 
 ethaddr=02:00:11:22:33:44
-eth3addr=02:00:11:22:33:45
-eth4addr=02:00:11:22:33:48
-eth5addr=02:00:11:22:33:46
 eth6addr=02:00:11:22:33:47
-eth8addr=02:00:11:22:33:49
 ipaddr=192.0.2.1
 
 /*
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (5 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 06/16] sandbox: Move some mac addresses to device tree Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-18 14:29   ` Vladimir Oltean
  2022-05-05 17:11 ` [PATCH v4 08/16] misc: i2c_eeprom: Make i2c_eeprom_write use a const buf Sean Anderson
                   ` (8 subsequent siblings)
  15 siblings, 1 reply; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson, Alex Marginean, Claudiu Manoil,
	Vladimir Oltean

If the DSA master fails to probe for whatever reason, then DSA devices
will continue on as if nothing is wrong. This can cause incorrect
behavior. In particular, on sandbox, dsa_sandbox_probe attempts to
access the master's private data. This is only safe to do if the master
has been probed first. Fix this by probing the master after we look it
up, and bailing out if we get an error.

Fixes: fc054d563b ("net: Introduce DSA class for Ethernet switches")
Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

Changes in v4:
- New

 net/dsa-uclass.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/dsa-uclass.c b/net/dsa-uclass.c
index 9ff55a02fb..3bf4351c84 100644
--- a/net/dsa-uclass.c
+++ b/net/dsa-uclass.c
@@ -477,8 +477,10 @@ static int dsa_pre_probe(struct udevice *dev)
 		return -ENODEV;
 	}
 
-	uclass_find_device_by_ofnode(UCLASS_ETH, pdata->master_node,
-				     &priv->master_dev);
+	err = uclass_get_device_by_ofnode(UCLASS_ETH, pdata->master_node,
+					  &priv->master_dev);
+	if (err)
+		return err;
 
 	/* Simulate a probing event for the CPU port */
 	if (ops->port_probe) {
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 08/16] misc: i2c_eeprom: Make i2c_eeprom_write use a const buf
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (6 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 09/16] misc: i2c_eeprom: Add fallbacks Sean Anderson
                   ` (7 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

i2c_eeprom_ops->write uses a const buf, so use one for the wrapper
function as well.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 drivers/misc/i2c_eeprom.c | 3 ++-
 include/i2c_eeprom.h      | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/misc/i2c_eeprom.c b/drivers/misc/i2c_eeprom.c
index 89a450d0f8..4302e180ac 100644
--- a/drivers/misc/i2c_eeprom.c
+++ b/drivers/misc/i2c_eeprom.c
@@ -33,7 +33,8 @@ int i2c_eeprom_read(struct udevice *dev, int offset, uint8_t *buf, int size)
 	return ops->read(dev, offset, buf, size);
 }
 
-int i2c_eeprom_write(struct udevice *dev, int offset, uint8_t *buf, int size)
+int i2c_eeprom_write(struct udevice *dev, int offset, const uint8_t *buf,
+		     int size)
 {
 	const struct i2c_eeprom_ops *ops = device_get_ops(dev);
 
diff --git a/include/i2c_eeprom.h b/include/i2c_eeprom.h
index 3ad565684f..90fdb25232 100644
--- a/include/i2c_eeprom.h
+++ b/include/i2c_eeprom.h
@@ -42,7 +42,8 @@ int i2c_eeprom_read(struct udevice *dev, int offset, uint8_t *buf, int size);
  *
  * Return: 0 on success, -ve on failure
  */
-int i2c_eeprom_write(struct udevice *dev, int offset, uint8_t *buf, int size);
+int i2c_eeprom_write(struct udevice *dev, int offset, const uint8_t *buf,
+		     int size);
 
 /*
  * i2c_eeprom_size() - get size of I2C EEPROM chip
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 09/16] misc: i2c_eeprom: Add fallbacks
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (7 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 08/16] misc: i2c_eeprom: Make i2c_eeprom_write use a const buf Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 10/16] misc: Add support for nvmem cells Sean Anderson
                   ` (6 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

Add some fallback functions for when i2c_eeprom is disabled. This allows
code to reference i2c_eeprom_* functions without needing to check
whether support has been compiled in.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

Changes in v4:
- New

 include/i2c_eeprom.h | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/include/i2c_eeprom.h b/include/i2c_eeprom.h
index 90fdb25232..32dcb03497 100644
--- a/include/i2c_eeprom.h
+++ b/include/i2c_eeprom.h
@@ -6,6 +6,8 @@
 #ifndef __I2C_EEPROM
 #define __I2C_EEPROM
 
+struct udevice;
+
 struct i2c_eeprom_ops {
 	int (*read)(struct udevice *dev, int offset, uint8_t *buf, int size);
 	int (*write)(struct udevice *dev, int offset, const uint8_t *buf,
@@ -20,6 +22,7 @@ struct i2c_eeprom {
 	unsigned long size;
 };
 
+#if CONFIG_IS_ENABLED(I2C_EEPROM)
 /*
  * i2c_eeprom_read() - read bytes from an I2C EEPROM chip
  *
@@ -54,4 +57,25 @@ int i2c_eeprom_write(struct udevice *dev, int offset, const uint8_t *buf,
  */
 int i2c_eeprom_size(struct udevice *dev);
 
+#else /* !I2C_EEPROM */
+
+static inline int i2c_eeprom_read(struct udevice *dev, int offset, uint8_t *buf,
+				  int size)
+{
+	return -ENOSYS;
+}
+
+static inline int i2c_eeprom_write(struct udevice *dev, int offset,
+				   const uint8_t *buf, int size)
+{
+	return -ENOSYS;
+}
+
+static inline int i2c_eeprom_size(struct udevice *dev)
+{
+	return -ENOSYS;
+}
+
+#endif /* I2C_EEPROM */
+
 #endif
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 10/16] misc: Add support for nvmem cells
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (8 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 09/16] misc: i2c_eeprom: Add fallbacks Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 11/16] sandbox: Enable NVMEM Sean Anderson
                   ` (5 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This adds support for "nvmem cells" as seen in Linux. The nvmem device
class in Linux is used for various assorted ROMs and EEPROMs. In this
sense, it is similar to UCLASS_MISC, but also includes
UCLASS_I2C_EEPROM, UCLASS_RTC, and UCLASS_MTD. New drivers corresponding
to a Linux-style nvmem device should be implemented as one of the
previously-mentioned uclasses. The nvmem API acts as a compatibility
layer to adapt the (slightly different) APIs of these uclasses. It also
handles the lookup of nvmem cells.

While nvmem devices can be accessed directly, they are most often used
by reading/writing contiguous values called "cells". Cells typically
hold information like calibration, versions, or configuration (such as
mac addresses).

nvmem devices can specify "cells" in their device tree:

	qfprom: eeprom@700000 {
		#address-cells = <1>;
		#size-cells = <1>;
		reg = <0x00700000 0x100000>;

		/* ... */

		tsens_calibration: calib@404 {
			reg = <0x404 0x10>;
		};
	};

which can then be referenced like:

	tsens {
		/* ... */
		nvmem-cells = <&tsens_calibration>;
		nvmem-cell-names = "calibration";
	};

The tsens driver could then read the calibration value like:

	struct nvmem_cell cal_cell;
	u8 cal[16];
	nvmem_cell_get_by_name(dev, "calibration", &cal_cell);
	nvmem_cell_read(&cal_cell, cal, sizeof(cal));

Because nvmem devices are not all of the same uclass, supported uclasses
must register a nvmem_interface struct. This allows CONFIG_NVMEM to be
enabled without depending on specific uclasses. At the moment,
nvmem_interface is very bare-bones, and assumes that no initialization
is necessary. However, this could be amended in the future.

Although I2C_EEPROM and MISC are quite similar (and could likely be
unified), they present different read/write function signatures. To
abstract over this, NVMEM uses the same read/write signature as Linux.
In particular, short read/writes are not allowed, which is allowed by
MISC.

The functionality implemented by nvmem cells is very similar to that
provided by i2c_eeprom_partition. "fixed-partition"s for eeproms does
not seem to have made its way into Linux or into any device tree other
than sandbox. It is possible that with the introduction of this API it
would be possible to remove it.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

Changes in v4:
- Fix failing to find nvmem devices if some supported uclasses were not
  enabled.
- Remove typedefs from documentation
- Document some of the design background, and why a new uclass was not
  created.

Changes in v2:
- Call the appropriate API functions directly from
  nvmem_cell_(read|write). This means we can drop the nvmem_interface
  machinery.

 MAINTAINERS           |   7 +++
 doc/api/index.rst     |   1 +
 doc/api/nvmem.rst     |  10 +++
 drivers/misc/Kconfig  |  16 +++++
 drivers/misc/Makefile |   1 +
 drivers/misc/nvmem.c  | 142 ++++++++++++++++++++++++++++++++++++++++++
 include/nvmem.h       | 134 +++++++++++++++++++++++++++++++++++++++
 7 files changed, 311 insertions(+)
 create mode 100644 doc/api/nvmem.rst
 create mode 100644 drivers/misc/nvmem.c
 create mode 100644 include/nvmem.h

diff --git a/MAINTAINERS b/MAINTAINERS
index 56be0bfad0..01f4581d6a 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1080,6 +1080,13 @@ F:	cmd/nvme.c
 F:	include/nvme.h
 F:	doc/develop/driver-model/nvme.rst
 
+NVMEM
+M:	Sean Anderson <seanga2@gmail.com>
+S:	Maintained
+F:	doc/api/nvmem.rst
+F:	drivers/misc/nvmem.c
+F:	include/nvmem.h
+
 NXP C45 TJA11XX PHY DRIVER
 M:	Radu Pirea <radu-nicolae.pirea@oss.nxp.com>
 S:	Maintained
diff --git a/doc/api/index.rst b/doc/api/index.rst
index 72fea981b7..a9338cfef9 100644
--- a/doc/api/index.rst
+++ b/doc/api/index.rst
@@ -14,6 +14,7 @@ U-Boot API documentation
    linker_lists
    lmb
    logging
+   nvmem
    pinctrl
    rng
    sandbox
diff --git a/doc/api/nvmem.rst b/doc/api/nvmem.rst
new file mode 100644
index 0000000000..d923784652
--- /dev/null
+++ b/doc/api/nvmem.rst
@@ -0,0 +1,10 @@
+.. SPDX-License-Identifier: GPL-2.0+
+
+NVMEM API
+=========
+
+.. kernel-doc:: include/nvmem.h
+   :doc: Design
+
+.. kernel-doc:: include/nvmem.h
+   :internal:
diff --git a/drivers/misc/Kconfig b/drivers/misc/Kconfig
index dab7a99c68..eaf7bb63ba 100644
--- a/drivers/misc/Kconfig
+++ b/drivers/misc/Kconfig
@@ -31,6 +31,22 @@ config TPL_MISC
 	  set of generic read, write and ioctl methods may be used to
 	  access the device.
 
+config NVMEM
+	bool "NVMEM support"
+	help
+	  This adds support for a common interface to different types of
+	  non-volatile memory. Consumers can use nvmem-cells properties to look
+	  up hardware configuration data such as MAC addresses and calibration
+	  settings.
+
+config SPL_NVMEM
+	bool "NVMEM support in SPL"
+	help
+	  This adds support for a common interface to different types of
+	  non-volatile memory. Consumers can use nvmem-cells properties to look
+	  up hardware configuration data such as MAC addresses and calibration
+	  settings.
+
 config ALTERA_SYSID
 	bool "Altera Sysid support"
 	depends on MISC
diff --git a/drivers/misc/Makefile b/drivers/misc/Makefile
index 6150d01e88..03fad7a23f 100644
--- a/drivers/misc/Makefile
+++ b/drivers/misc/Makefile
@@ -4,6 +4,7 @@
 # Wolfgang Denk, DENX Software Engineering, wd@denx.de.
 
 obj-$(CONFIG_MISC) += misc-uclass.o
+obj-$(CONFIG_$(SPL_TPL_)NVMEM) += nvmem.o
 
 obj-$(CONFIG_$(SPL_TPL_)CROS_EC) += cros_ec.o
 obj-$(CONFIG_$(SPL_TPL_)CROS_EC_SANDBOX) += cros_ec_sandbox.o
diff --git a/drivers/misc/nvmem.c b/drivers/misc/nvmem.c
new file mode 100644
index 0000000000..5a2bd1f9f7
--- /dev/null
+++ b/drivers/misc/nvmem.c
@@ -0,0 +1,142 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2022 Sean Anderson <sean.anderson@seco.com>
+ */
+
+#include <common.h>
+#include <i2c_eeprom.h>
+#include <linker_lists.h>
+#include <misc.h>
+#include <nvmem.h>
+#include <rtc.h>
+#include <dm/device_compat.h>
+#include <dm/ofnode.h>
+#include <dm/read.h>
+#include <dm/uclass.h>
+
+int nvmem_cell_read(struct nvmem_cell *cell, void *buf, size_t size)
+{
+	dev_dbg(cell->nvmem, "%s: off=%u size=%zu\n", __func__, cell->offset, size);
+	if (size != cell->size)
+		return -EINVAL;
+
+	switch (cell->nvmem->driver->id) {
+	case UCLASS_I2C_EEPROM:
+		return i2c_eeprom_read(cell->nvmem, cell->offset, buf, size);
+	case UCLASS_MISC: {
+		int ret = misc_read(cell->nvmem, cell->offset, buf, size);
+
+		if (ret < 0)
+			return ret;
+		if (ret != size)
+			return -EIO;
+		return 0;
+	}
+	case UCLASS_RTC:
+		return dm_rtc_read(cell->nvmem, cell->offset, buf, size);
+	default:
+		return -ENOSYS;
+	}
+}
+
+int nvmem_cell_write(struct nvmem_cell *cell, const void *buf, size_t size)
+{
+	dev_dbg(cell->nvmem, "%s: off=%u size=%zu\n", __func__, cell->offset, size);
+	if (size != cell->size)
+		return -EINVAL;
+
+	switch (cell->nvmem->driver->id) {
+	case UCLASS_I2C_EEPROM:
+		return i2c_eeprom_write(cell->nvmem, cell->offset, buf, size);
+	case UCLASS_MISC: {
+		int ret = misc_write(cell->nvmem, cell->offset, buf, size);
+
+		if (ret < 0)
+			return ret;
+		if (ret != size)
+			return -EIO;
+		return 0;
+	}
+	case UCLASS_RTC:
+		return dm_rtc_write(cell->nvmem, cell->offset, buf, size);
+	default:
+		return -ENOSYS;
+	}
+}
+
+/**
+ * nvmem_get_device() - Get an nvmem device for a cell
+ * @node: ofnode of the nvmem device
+ * @cell: Cell to look up
+ *
+ * Try to find a nvmem-compatible device by going through the nvmem interfaces.
+ *
+ * Return:
+ * * 0 on success
+ * * -ENODEV if we didn't find anything
+ * * A negative error if there was a problem looking up the device
+ */
+static int nvmem_get_device(ofnode node, struct nvmem_cell *cell)
+{
+	int i, ret;
+	enum uclass_id ids[] = {
+		UCLASS_I2C_EEPROM,
+		UCLASS_MISC,
+		UCLASS_RTC,
+	};
+
+	for (i = 0; i < ARRAY_SIZE(ids); i++) {
+		ret = uclass_get_device_by_ofnode(ids[i], node, &cell->nvmem);
+		if (!ret)
+			return 0;
+		if (ret != -ENODEV && ret != -EPFNOSUPPORT)
+			return ret;
+	}
+
+	return -ENODEV;
+}
+
+int nvmem_cell_get_by_index(struct udevice *dev, int index,
+			    struct nvmem_cell *cell)
+{
+	fdt_addr_t offset;
+	fdt_size_t size = FDT_SIZE_T_NONE;
+	int ret;
+	struct ofnode_phandle_args args;
+
+	dev_dbg(dev, "%s: index=%d\n", __func__, index);
+
+	ret = dev_read_phandle_with_args(dev, "nvmem-cells", NULL, 0, index,
+					 &args);
+	if (ret)
+		return ret;
+
+	ret = nvmem_get_device(ofnode_get_parent(args.node), cell);
+	if (ret)
+		return ret;
+
+	offset = ofnode_get_addr_size_index_notrans(args.node, 0, &size);
+	if (offset == FDT_ADDR_T_NONE || size == FDT_SIZE_T_NONE) {
+		dev_dbg(cell->nvmem, "missing address or size for %s\n",
+			ofnode_get_name(args.node));
+		return -EINVAL;
+	}
+
+	cell->offset = offset;
+	cell->size = size;
+	return 0;
+}
+
+int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
+			   struct nvmem_cell *cell)
+{
+	int index;
+
+	dev_dbg(dev, "%s, name=%s\n", __func__, name);
+
+	index = dev_read_stringlist_search(dev, "nvmem-cell-names", name);
+	if (index < 0)
+		return index;
+
+	return nvmem_cell_get_by_index(dev, index, cell);
+}
diff --git a/include/nvmem.h b/include/nvmem.h
new file mode 100644
index 0000000000..822e698bdd
--- /dev/null
+++ b/include/nvmem.h
@@ -0,0 +1,134 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (c) 2022 Sean Anderson <sean.anderson@seco.com>
+ */
+
+#ifndef NVMEM_H
+#define NVMEM_H
+
+/**
+ * DOC: Design
+ *
+ * The NVMEM subsystem is a "meta-uclass" in that it abstracts over several
+ * different uclasses all with read/write APIs. One approach to implementing
+ * this could be to add a new sub-device for each nvmem-style device of
+ * UCLASS_NVMEM. This subsystem has taken the approach of using the existing
+ * access methods (i2c_eeprom_write, misc_write, etc.) directly. This has the
+ * advantage of not requiring an extra device/driver, saving on binary size and
+ * runtime memory usage. On the other hand, it is not idiomatic. Similar
+ * efforts should generally use a new uclass.
+ */
+
+/**
+ * struct nvmem_cell - One datum within non-volatile memory
+ * @nvmem: The backing storage device
+ * @offset: The offset of the cell from the start of @nvmem
+ * @size: The size of the cell, in bytes
+ */
+struct nvmem_cell {
+	struct udevice *nvmem;
+	unsigned int offset;
+	size_t size;
+};
+
+struct udevice;
+
+#if CONFIG_IS_ENABLED(NVMEM)
+
+/**
+ * nvmem_cell_read() - Read the value of an nvmem cell
+ * @cell: The nvmem cell to read
+ * @buf: The buffer to read into
+ * @size: The size of @buf
+ *
+ * Return:
+ * * 0 on success
+ * * -EINVAL if @buf is not the same size as @cell.
+ * * -ENOSYS if CONFIG_NVMEM is disabled
+ * * A negative error if there was a problem reading the underlying storage
+ */
+int nvmem_cell_read(struct nvmem_cell *cell, void *buf, size_t size);
+
+/**
+ * nvmem_cell_write() - Write a value to an nvmem cell
+ * @cell: The nvmem cell to write
+ * @buf: The buffer to write from
+ * @size: The size of @buf
+ *
+ * Return:
+ * * 0 on success
+ * * -EINVAL if @buf is not the same size as @cell
+ * * -ENOSYS if @cell is read-only, or if CONFIG_NVMEM is disabled
+ * * A negative error if there was a problem writing the underlying storage
+ */
+int nvmem_cell_write(struct nvmem_cell *cell, const void *buf, size_t size);
+
+/**
+ * nvmem_cell_get_by_index() - Get an nvmem cell from a given device and index
+ * @dev: The device that uses the nvmem cell
+ * @index: The index of the cell in nvmem-cells
+ * @cell: The cell to initialize
+ *
+ * Look up the nvmem cell referenced by the phandle at @index in nvmem-cells in
+ * @dev.
+ *
+ * Return:
+ * * 0 on success
+ * * -EINVAL if the regs property is missing, empty, or undersized
+ * * -ENODEV if the nvmem device is missing or unimplemented
+ * * -ENOSYS if CONFIG_NVMEM is disabled
+ * * A negative error if there was a problem reading nvmem-cells or getting the
+ *   device
+ */
+int nvmem_cell_get_by_index(struct udevice *dev, int index,
+			    struct nvmem_cell *cell);
+
+/**
+ * nvmem_cell_get_by_name() - Get an nvmem cell from a given device and name
+ * @dev: The device that uses the nvmem cell
+ * @name: The name of the nvmem cell
+ * @cell: The cell to initialize
+ *
+ * Look up the nvmem cell referenced by @name in the nvmem-cell-names property
+ * of @dev.
+ *
+ * Return:
+ * * 0 on success
+ * * -EINVAL if the regs property is missing, empty, or undersized
+ * * -ENODEV if the nvmem device is missing or unimplemented
+ * * -ENODATA if @name is not in nvmem-cell-names
+ * * -ENOSYS if CONFIG_NVMEM is disabled
+ * * A negative error if there was a problem reading nvmem-cell-names,
+ *   nvmem-cells, or getting the device
+ */
+int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
+			   struct nvmem_cell *cell);
+
+#else /* CONFIG_NVMEM */
+
+static inline int nvmem_cell_read(struct nvmem_cell *cell, void *buf, int size)
+{
+	return -ENOSYS;
+}
+
+static inline int nvmem_cell_write(struct nvmem_cell *cell, const void *buf,
+				   int size)
+{
+	return -ENOSYS;
+}
+
+static inline int nvmem_cell_get_by_index(struct udevice *dev, int index,
+					  struct nvmem_cell *cell)
+{
+	return -ENOSYS;
+}
+
+static inline int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
+					 struct nvmem_cell *cell)
+{
+	return -ENOSYS;
+}
+
+#endif /* CONFIG_NVMEM */
+
+#endif /* NVMEM_H */
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 11/16] sandbox: Enable NVMEM
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (9 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 10/16] misc: Add support for nvmem cells Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 12/16] net: Add support for reading mac addresses from nvmem cells Sean Anderson
                   ` (4 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This enables NVMEM for all sandbox defconfigs, enabling it to be used in
unit tests in the next few commits.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

Changes in v4:
- Enable CONFIG_NVMEM for sandbox_spl_defconfig

 configs/sandbox64_defconfig        | 1 +
 configs/sandbox_defconfig          | 1 +
 configs/sandbox_flattree_defconfig | 1 +
 configs/sandbox_noinst_defconfig   | 1 +
 configs/sandbox_spl_defconfig      | 2 ++
 5 files changed, 6 insertions(+)

diff --git a/configs/sandbox64_defconfig b/configs/sandbox64_defconfig
index d7f22b39ae..fc7e100eaa 100644
--- a/configs/sandbox64_defconfig
+++ b/configs/sandbox64_defconfig
@@ -144,6 +144,7 @@ CONFIG_LED_GPIO=y
 CONFIG_DM_MAILBOX=y
 CONFIG_SANDBOX_MBOX=y
 CONFIG_MISC=y
+CONFIG_NVMEM=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_I2C=y
 CONFIG_CROS_EC_LPC=y
diff --git a/configs/sandbox_defconfig b/configs/sandbox_defconfig
index fe8ea4626d..cf02027b12 100644
--- a/configs/sandbox_defconfig
+++ b/configs/sandbox_defconfig
@@ -187,6 +187,7 @@ CONFIG_LED_GPIO=y
 CONFIG_DM_MAILBOX=y
 CONFIG_SANDBOX_MBOX=y
 CONFIG_MISC=y
+CONFIG_NVMEM=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_I2C=y
 CONFIG_CROS_EC_LPC=y
diff --git a/configs/sandbox_flattree_defconfig b/configs/sandbox_flattree_defconfig
index 80a4be47eb..6eb4e348a2 100644
--- a/configs/sandbox_flattree_defconfig
+++ b/configs/sandbox_flattree_defconfig
@@ -117,6 +117,7 @@ CONFIG_LED_GPIO=y
 CONFIG_DM_MAILBOX=y
 CONFIG_SANDBOX_MBOX=y
 CONFIG_MISC=y
+CONFIG_NVMEM=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_I2C=y
 CONFIG_CROS_EC_LPC=y
diff --git a/configs/sandbox_noinst_defconfig b/configs/sandbox_noinst_defconfig
index c9430da0f0..e9cb299538 100644
--- a/configs/sandbox_noinst_defconfig
+++ b/configs/sandbox_noinst_defconfig
@@ -143,6 +143,7 @@ CONFIG_LED_GPIO=y
 CONFIG_DM_MAILBOX=y
 CONFIG_SANDBOX_MBOX=y
 CONFIG_MISC=y
+CONFIG_NVMEM=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_I2C=y
 CONFIG_CROS_EC_LPC=y
diff --git a/configs/sandbox_spl_defconfig b/configs/sandbox_spl_defconfig
index 13a76e89ea..0a32311f77 100644
--- a/configs/sandbox_spl_defconfig
+++ b/configs/sandbox_spl_defconfig
@@ -144,6 +144,8 @@ CONFIG_LED_GPIO=y
 CONFIG_DM_MAILBOX=y
 CONFIG_SANDBOX_MBOX=y
 CONFIG_MISC=y
+CONFIG_NVMEM=y
+CONFIG_SPL_NVMEM=y
 CONFIG_CROS_EC=y
 CONFIG_CROS_EC_I2C=y
 CONFIG_CROS_EC_LPC=y
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 12/16] net: Add support for reading mac addresses from nvmem cells
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (10 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 11/16] sandbox: Enable NVMEM Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 13/16] test: Load mac address with i2c eeprom Sean Anderson
                   ` (3 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This adds support for reading mac addresses from the "mac-address" nvmem
cell. If there is no (local-)mac-address property, then we will try
reading from an nvmem cell.

For some existing examples of this property, refer to imx8mn.dtsi and
imx8mp.dtsi. Unfortunately, fuse drivers have not yet been converted
to DM.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 net/eth-uclass.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/net/eth-uclass.c b/net/eth-uclass.c
index bcefc54ded..0f6b45b002 100644
--- a/net/eth-uclass.c
+++ b/net/eth-uclass.c
@@ -14,6 +14,7 @@
 #include <env.h>
 #include <log.h>
 #include <net.h>
+#include <nvmem.h>
 #include <asm/global_data.h>
 #include <dm/device-internal.h>
 #include <dm/uclass-internal.h>
@@ -507,17 +508,21 @@ static bool eth_dev_get_mac_address(struct udevice *dev, u8 mac[ARP_HLEN])
 {
 #if CONFIG_IS_ENABLED(OF_CONTROL)
 	const uint8_t *p;
+	struct nvmem_cell mac_cell;
 
 	p = dev_read_u8_array_ptr(dev, "mac-address", ARP_HLEN);
 	if (!p)
 		p = dev_read_u8_array_ptr(dev, "local-mac-address", ARP_HLEN);
 
-	if (!p)
+	if (p) {
+		memcpy(mac, p, ARP_HLEN);
+		return true;
+	}
+
+	if (nvmem_cell_get_by_name(dev, "mac-address", &mac_cell))
 		return false;
 
-	memcpy(mac, p, ARP_HLEN);
-
-	return true;
+	return !nvmem_cell_read(&mac_cell, mac, ARP_HLEN);
 #else
 	return false;
 #endif
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 13/16] test: Load mac address with i2c eeprom
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (11 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 12/16] net: Add support for reading mac addresses from nvmem cells Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 14/16] test: Load mac address using RTC Sean Anderson
                   ` (2 subsequent siblings)
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This uses an i2c eeprom to load a mac address using the nvmem interface.
Enable I2C_EEPROM for sandbox SPL since it is the only sandbox config
which doesn't enable it eeprom.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

Changes in v4:
- Enable CONFIG_I2C_EEPROM for sandbox_spl_defconfig

 arch/sandbox/dts/test.dts      | 9 ++++++++-
 configs/sandbox_spl_defconfig  | 1 +
 drivers/misc/i2c_eeprom_emul.c | 4 ++++
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index 318dc2dcb1..9aed071aa0 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -538,7 +538,8 @@
 	eth_3: sbe5 {
 		compatible = "sandbox,eth";
 		reg = <0x10005000 0x1000>;
-		mac-address = [ 02 00 11 22 33 45 ];
+		nvmem-cells = <&eth3_addr>;
+		nvmem-cell-names = "mac-address";
 	};
 
 	eth@10004000 {
@@ -701,6 +702,8 @@
 		pinctrl-0 = <&pinmux_i2c0_pins>;
 
 		eeprom@2c {
+			#address-cells = <1>;
+			#size-cells = <1>;
 			reg = <0x2c>;
 			compatible = "i2c-eeprom";
 			sandbox,emul = <&emul_eeprom>;
@@ -712,6 +715,10 @@
 					reg = <10 2>;
 				};
 			};
+
+			eth3_addr: mac-address@24 {
+				reg = <24 6>;
+			};
 		};
 
 		rtc_0: rtc@43 {
diff --git a/configs/sandbox_spl_defconfig b/configs/sandbox_spl_defconfig
index 0a32311f77..78d295f1f8 100644
--- a/configs/sandbox_spl_defconfig
+++ b/configs/sandbox_spl_defconfig
@@ -154,6 +154,7 @@ CONFIG_CROS_EC_SPI=y
 CONFIG_P2SB=y
 CONFIG_PWRSEQ=y
 CONFIG_SPL_PWRSEQ=y
+CONFIG_I2C_EEPROM=y
 CONFIG_MMC_SANDBOX=y
 CONFIG_SPI_FLASH_SANDBOX=y
 CONFIG_SPI_FLASH_ATMEL=y
diff --git a/drivers/misc/i2c_eeprom_emul.c b/drivers/misc/i2c_eeprom_emul.c
index 85b127c406..6f32087ede 100644
--- a/drivers/misc/i2c_eeprom_emul.c
+++ b/drivers/misc/i2c_eeprom_emul.c
@@ -171,11 +171,15 @@ static int sandbox_i2c_eeprom_probe(struct udevice *dev)
 {
 	struct sandbox_i2c_flash_plat_data *plat = dev_get_plat(dev);
 	struct sandbox_i2c_flash *priv = dev_get_priv(dev);
+	/* For eth3 */
+	const u8 mac[] = { 0x02, 0x00, 0x11, 0x22, 0x33, 0x45 };
 
 	priv->data = calloc(1, plat->size);
 	if (!priv->data)
 		return -ENOMEM;
 
+	memcpy(&priv->data[24], mac, sizeof(mac));
+
 	return 0;
 }
 
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 14/16] test: Load mac address using RTC
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (12 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 13/16] test: Load mac address with i2c eeprom Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 15/16] test: Load mac address using misc device Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 16/16] [RFC] misc: nvmem: Convert to using udevices Sean Anderson
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This uses the nvmem API to load a mac address from an RTC.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 arch/sandbox/dts/test.dts  |  9 ++++++++-
 drivers/rtc/i2c_rtc_emul.c | 10 ++++++++++
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index 9aed071aa0..3e0b454efb 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -558,7 +558,8 @@
 	dsa_eth0: dsa-test-eth {
 		compatible = "sandbox,eth";
 		reg = <0x10006000 0x1000>;
-		mac-address = [ 02 00 11 22 33 48 ];
+		nvmem-cells = <&eth4_addr>;
+		nvmem-cell-names = "mac-address";
 	};
 
 	dsa-test {
@@ -722,9 +723,15 @@
 		};
 
 		rtc_0: rtc@43 {
+			#address-cells = <1>;
+			#size-cells = <1>;
 			reg = <0x43>;
 			compatible = "sandbox-rtc";
 			sandbox,emul = <&emul0>;
+
+			eth4_addr: mac-address@40 {
+				reg = <0x40 6>;
+			};
 		};
 
 		rtc_1: rtc@61 {
diff --git a/drivers/rtc/i2c_rtc_emul.c b/drivers/rtc/i2c_rtc_emul.c
index ba418c25da..c307d6036d 100644
--- a/drivers/rtc/i2c_rtc_emul.c
+++ b/drivers/rtc/i2c_rtc_emul.c
@@ -203,6 +203,15 @@ static int sandbox_i2c_rtc_bind(struct udevice *dev)
 	return 0;
 }
 
+static int sandbox_i2c_rtc_probe(struct udevice *dev)
+{
+	const u8 mac[] = { 0x02, 0x00, 0x11, 0x22, 0x33, 0x48 };
+	struct sandbox_i2c_rtc_plat_data *plat = dev_get_plat(dev);
+
+	memcpy(&plat->reg[0x40], mac, sizeof(mac));
+	return 0;
+}
+
 static const struct udevice_id sandbox_i2c_rtc_ids[] = {
 	{ .compatible = "sandbox,i2c-rtc-emul" },
 	{ }
@@ -213,6 +222,7 @@ U_BOOT_DRIVER(sandbox_i2c_rtc_emul) = {
 	.id		= UCLASS_I2C_EMUL,
 	.of_match	= sandbox_i2c_rtc_ids,
 	.bind		= sandbox_i2c_rtc_bind,
+	.probe		= sandbox_i2c_rtc_probe,
 	.priv_auto	= sizeof(struct sandbox_i2c_rtc),
 	.plat_auto	= sizeof(struct sandbox_i2c_rtc_plat_data),
 	.ops		= &sandbox_i2c_rtc_emul_ops,
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 15/16] test: Load mac address using misc device
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (13 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 14/16] test: Load mac address using RTC Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  2022-05-05 17:11 ` [PATCH v4 16/16] [RFC] misc: nvmem: Convert to using udevices Sean Anderson
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

This loads a mac address using a misc device using the nvmem interface.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
---

(no changes since v1)

 arch/sandbox/dts/test.dts   | 9 ++++++++-
 drivers/misc/misc_sandbox.c | 3 +++
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index 3e0b454efb..c77eaf4950 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -532,7 +532,8 @@
 	eth_5: eth@10003000 {
 		compatible = "sandbox,eth";
 		reg = <0x10003000 0x1000>;
-		mac-address = [ 02 00 11 22 33 46 ];
+		nvmem-cells = <&eth5_addr>;
+		nvmem-cell-names = "mac-address";
 	};
 
 	eth_3: sbe5 {
@@ -913,7 +914,13 @@
 	};
 
 	misc-test {
+		#address-cells = <1>;
+		#size-cells = <1>;
 		compatible = "sandbox,misc_sandbox";
+
+		eth5_addr: mac-address@10 {
+			reg = <0x10 6>;
+		};
 	};
 
 	mmc2 {
diff --git a/drivers/misc/misc_sandbox.c b/drivers/misc/misc_sandbox.c
index 0e4292fd0a..31cde2dbac 100644
--- a/drivers/misc/misc_sandbox.c
+++ b/drivers/misc/misc_sandbox.c
@@ -112,8 +112,11 @@ static const struct misc_ops misc_sandbox_ops = {
 int misc_sandbox_probe(struct udevice *dev)
 {
 	struct misc_sandbox_priv *priv = dev_get_priv(dev);
+	/* For eth5 */
+	const u8 mac[] = { 0x02, 0x00, 0x11, 0x22, 0x33, 0x46 };
 
 	priv->enabled = true;
+	memcpy(&priv->mem[16], mac, sizeof(mac));
 
 	return 0;
 }
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* [PATCH v4 16/16] [RFC] misc: nvmem: Convert to using udevices
  2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
                   ` (14 preceding siblings ...)
  2022-05-05 17:11 ` [PATCH v4 15/16] test: Load mac address using misc device Sean Anderson
@ 2022-05-05 17:11 ` Sean Anderson
  15 siblings, 0 replies; 19+ messages in thread
From: Sean Anderson @ 2022-05-05 17:11 UTC (permalink / raw)
  To: u-boot, Simon Glass
  Cc: Heinrich Schuchardt, Mario Six, Ramon Fried, Tom Rini,
	Joe Hershberger, Sean Anderson

Instead of calling uclass methods directly, instead create some nvmem
devices solely for the purpose of holding nvmem ops. This is primarily
to illustrate the size difference between these approaches.

This patch should not be applied.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Patch-prefix: RFC
---

Changes in v4:
- New

 doc/api/nvmem.rst          |   3 --
 drivers/misc/i2c_eeprom.c  |  37 ++++++++++++++
 drivers/misc/misc-uclass.c |  58 +++++++++++++++++++--
 drivers/misc/nvmem.c       | 100 ++++++++++++-------------------------
 drivers/rtc/rtc-uclass.c   |  46 +++++++++++++++--
 include/dm/uclass-id.h     |   1 +
 include/nvmem.h            |  79 +++++++++++++++++++++++------
 7 files changed, 233 insertions(+), 91 deletions(-)

diff --git a/doc/api/nvmem.rst b/doc/api/nvmem.rst
index d923784652..15c9b5b839 100644
--- a/doc/api/nvmem.rst
+++ b/doc/api/nvmem.rst
@@ -3,8 +3,5 @@
 NVMEM API
 =========
 
-.. kernel-doc:: include/nvmem.h
-   :doc: Design
-
 .. kernel-doc:: include/nvmem.h
    :internal:
diff --git a/drivers/misc/i2c_eeprom.c b/drivers/misc/i2c_eeprom.c
index 4302e180ac..3f6c7ebf4a 100644
--- a/drivers/misc/i2c_eeprom.c
+++ b/drivers/misc/i2c_eeprom.c
@@ -14,6 +14,7 @@
 #include <dm/device-internal.h>
 #include <i2c.h>
 #include <i2c_eeprom.h>
+#include <nvmem.h>
 
 struct i2c_eeprom_drv_data {
 	u32 size; /* size in bytes */
@@ -374,7 +375,43 @@ U_BOOT_DRIVER(i2c_eeprom_partition) = {
 	.ops			= &i2c_eeprom_partition_ops,
 };
 
+static int i2c_eeprom_nvmem_read(struct udevice *dev, unsigned int offset,
+				 void *buf, size_t size)
+{
+	return i2c_eeprom_read(dev_get_parent(dev), offset, buf, size);
+}
+
+static int i2c_eeprom_nvmem_write(struct udevice *dev, unsigned int offset,
+				  const void *buf, size_t size)
+{
+	return i2c_eeprom_write(dev_get_parent(dev), offset, buf, size);
+}
+
+static struct __maybe_unused nvmem_ops i2c_eeprom_nvmem_ops = {
+	.read = i2c_eeprom_nvmem_read,
+	.write = i2c_eeprom_nvmem_write,
+};
+
+#if CONFIG_IS_ENABLED(NVMEM)
+U_BOOT_DRIVER(i2c_eeprom_nvmem) = {
+	.name	= "i2c_eeprom_nvmem",
+	.id	= UCLASS_NVMEM,
+	.ops	= &i2c_eeprom_nvmem_ops,
+	.flags	= DM_FLAG_NAME_ALLOCED,
+};
+#endif
+
+#if CONFIG_IS_ENABLED(NVMEM)
+static int i2c_eeprom_post_bind(struct udevice *dev)
+{
+	return nvmem_register(dev, DM_DRIVER_GET(misc_nvmem));
+}
+#endif
+
 UCLASS_DRIVER(i2c_eeprom) = {
 	.id		= UCLASS_I2C_EEPROM,
 	.name		= "i2c_eeprom",
+#if CONFIG_IS_ENABLED(NVMEM)
+	.post_bind	= i2c_eeprom_post_bind,
+#endif
 };
diff --git a/drivers/misc/misc-uclass.c b/drivers/misc/misc-uclass.c
index cfe9d562fa..e9b58d3abe 100644
--- a/drivers/misc/misc-uclass.c
+++ b/drivers/misc/misc-uclass.c
@@ -9,6 +9,7 @@
 #include <dm.h>
 #include <errno.h>
 #include <misc.h>
+#include <nvmem.h>
 
 /*
  * Implement a  miscellaneous uclass for those do not fit other more
@@ -67,10 +68,61 @@ int misc_set_enabled(struct udevice *dev, bool val)
 	return ops->set_enabled(dev, val);
 }
 
+static int misc_nvmem_read(struct udevice *dev, unsigned int offset, void *buf,
+			   size_t size)
+{
+	int ret = misc_read(dev_get_parent(dev), offset, buf, size);
+
+	if (ret < 0)
+		return ret;
+	if (ret != size)
+		return -EIO;
+	return 0;
+}
+
+static int misc_nvmem_write(struct udevice *dev, unsigned int offset,
+			    const void *buf, size_t size)
+{
+	int ret = misc_write(dev_get_parent(dev), offset, buf, size);
+
+	if (ret < 0)
+		return ret;
+	if (ret != size)
+		return -EIO;
+	return 0;
+}
+
+static struct __maybe_unused nvmem_ops misc_nvmem_ops = {
+	.read = misc_nvmem_read,
+	.write = misc_nvmem_write,
+};
+
+#if CONFIG_IS_ENABLED(NVMEM)
+U_BOOT_DRIVER(misc_nvmem) = {
+	.name	= "misc_nvmem",
+	.id	= UCLASS_NVMEM,
+	.ops	= &misc_nvmem_ops,
+	.flags	= DM_FLAG_NAME_ALLOCED,
+};
+#endif
+
+static int misc_post_bind(struct udevice *dev)
+{
+#if CONFIG_IS_ENABLED(OF_REAL)
+	int ret = dm_scan_fdt_dev(dev);
+
+	if (ret)
+		return ret;
+#endif
+#if CONFIG_IS_ENABLED(NVMEM)
+	return nvmem_register(dev, DM_DRIVER_GET(misc_nvmem));
+#else
+	return 0;
+#endif
+}
+
 UCLASS_DRIVER(misc) = {
 	.id		= UCLASS_MISC,
 	.name		= "misc",
-#if CONFIG_IS_ENABLED(OF_REAL)
-	.post_bind	= dm_scan_fdt_dev,
-#endif
+	.post_bind	= misc_post_bind,
 };
diff --git a/drivers/misc/nvmem.c b/drivers/misc/nvmem.c
index 5a2bd1f9f7..afd8c7b5ab 100644
--- a/drivers/misc/nvmem.c
+++ b/drivers/misc/nvmem.c
@@ -10,90 +10,48 @@
 #include <nvmem.h>
 #include <rtc.h>
 #include <dm/device_compat.h>
+#include <dm/device-internal.h>
 #include <dm/ofnode.h>
 #include <dm/read.h>
 #include <dm/uclass.h>
 
+int nvmem_register(struct udevice *dev, const struct driver* drv)
+{
+	char *name;
+	int name_size;
+	static const char fmt[] = "%s.nvmem";
+
+	assert(drv->flags & DM_FLAG_NAME_ALLOCED);
+
+	name_size = snprintf(NULL, 0, fmt, dev->name) + 1;
+	name = malloc(name_size);
+	if (!name)
+		return -ENOMEM;
+	snprintf(name, name_size, fmt, dev->name);
+
+	return device_bind(dev, drv, name, NULL, dev_ofnode(dev), NULL);
+}
+
 int nvmem_cell_read(struct nvmem_cell *cell, void *buf, size_t size)
-{
+{	
+	const struct nvmem_ops *ops = dev_get_driver_ops(cell->nvmem);
+
 	dev_dbg(cell->nvmem, "%s: off=%u size=%zu\n", __func__, cell->offset, size);
 	if (size != cell->size)
 		return -EINVAL;
 
-	switch (cell->nvmem->driver->id) {
-	case UCLASS_I2C_EEPROM:
-		return i2c_eeprom_read(cell->nvmem, cell->offset, buf, size);
-	case UCLASS_MISC: {
-		int ret = misc_read(cell->nvmem, cell->offset, buf, size);
-
-		if (ret < 0)
-			return ret;
-		if (ret != size)
-			return -EIO;
-		return 0;
-	}
-	case UCLASS_RTC:
-		return dm_rtc_read(cell->nvmem, cell->offset, buf, size);
-	default:
-		return -ENOSYS;
-	}
+	return ops->read(cell->nvmem, cell->offset, buf, size);
 }
 
 int nvmem_cell_write(struct nvmem_cell *cell, const void *buf, size_t size)
 {
+	const struct nvmem_ops *ops = dev_get_driver_ops(cell->nvmem);
+
 	dev_dbg(cell->nvmem, "%s: off=%u size=%zu\n", __func__, cell->offset, size);
 	if (size != cell->size)
 		return -EINVAL;
 
-	switch (cell->nvmem->driver->id) {
-	case UCLASS_I2C_EEPROM:
-		return i2c_eeprom_write(cell->nvmem, cell->offset, buf, size);
-	case UCLASS_MISC: {
-		int ret = misc_write(cell->nvmem, cell->offset, buf, size);
-
-		if (ret < 0)
-			return ret;
-		if (ret != size)
-			return -EIO;
-		return 0;
-	}
-	case UCLASS_RTC:
-		return dm_rtc_write(cell->nvmem, cell->offset, buf, size);
-	default:
-		return -ENOSYS;
-	}
-}
-
-/**
- * nvmem_get_device() - Get an nvmem device for a cell
- * @node: ofnode of the nvmem device
- * @cell: Cell to look up
- *
- * Try to find a nvmem-compatible device by going through the nvmem interfaces.
- *
- * Return:
- * * 0 on success
- * * -ENODEV if we didn't find anything
- * * A negative error if there was a problem looking up the device
- */
-static int nvmem_get_device(ofnode node, struct nvmem_cell *cell)
-{
-	int i, ret;
-	enum uclass_id ids[] = {
-		UCLASS_I2C_EEPROM,
-		UCLASS_MISC,
-		UCLASS_RTC,
-	};
-
-	for (i = 0; i < ARRAY_SIZE(ids); i++) {
-		ret = uclass_get_device_by_ofnode(ids[i], node, &cell->nvmem);
-		if (!ret)
-			return 0;
-		if (ret != -ENODEV && ret != -EPFNOSUPPORT)
-			return ret;
-	}
-
-	return -ENODEV;
+	return ops->write(cell->nvmem, cell->offset, buf, size);
 }
 
 int nvmem_cell_get_by_index(struct udevice *dev, int index,
@@ -111,7 +69,8 @@ int nvmem_cell_get_by_index(struct udevice *dev, int index,
 	if (ret)
 		return ret;
 
-	ret = nvmem_get_device(ofnode_get_parent(args.node), cell);
+	ret = uclass_get_device_by_ofnode(UCLASS_NVMEM, ofnode_get_parent(args.node),
+					  &cell->nvmem);
 	if (ret)
 		return ret;
 
@@ -140,3 +99,8 @@ int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
 
 	return nvmem_cell_get_by_index(dev, index, cell);
 }
+
+UCLASS_DRIVER(nvmem) = {
+	.id		= UCLASS_NVMEM,
+	.name		= "nvmem",
+};
diff --git a/drivers/rtc/rtc-uclass.c b/drivers/rtc/rtc-uclass.c
index e5ae6ea4d5..bef2eefecf 100644
--- a/drivers/rtc/rtc-uclass.c
+++ b/drivers/rtc/rtc-uclass.c
@@ -10,6 +10,7 @@
 #include <dm.h>
 #include <errno.h>
 #include <log.h>
+#include <nvmem.h>
 #include <rtc.h>
 
 int dm_rtc_get(struct udevice *dev, struct rtc_time *time)
@@ -173,11 +174,50 @@ int rtc_write32(struct udevice *dev, unsigned int reg, u32 value)
 	return 0;
 }
 
+static int nvmem_rtc_read(struct udevice *dev, unsigned int offset, void *buf,
+			  size_t size)
+{
+	return dm_rtc_read(dev_get_parent(dev), offset, buf, size);
+}
+
+static int nvmem_rtc_write(struct udevice *dev, unsigned int offset,
+			   const void *buf, size_t size)
+{
+	return dm_rtc_write(dev_get_parent(dev), offset, buf, size);
+}
+
+static struct __maybe_unused nvmem_ops rtc_nvmem_ops = {
+	.read = nvmem_rtc_read,
+	.write = nvmem_rtc_write,
+};
+
+#if CONFIG_IS_ENABLED(NVMEM)
+U_BOOT_DRIVER(rtc_nvmem) = {
+	.name	= "rtc_nvmem",
+	.id	= UCLASS_NVMEM,
+	.ops	= &rtc_nvmem_ops,
+	.flags	= DM_FLAG_NAME_ALLOCED,
+};
+#endif
+
+static int rtc_post_bind(struct udevice *dev)
+{
+#if CONFIG_IS_ENABLED(OF_REAL)
+	int ret = dm_scan_fdt_dev(dev);
+
+	if (ret)
+		return ret;
+#endif
+#if CONFIG_IS_ENABLED(NVMEM)
+	return nvmem_register(dev, DM_DRIVER_GET(rtc_nvmem));
+#else
+	return 0;
+#endif
+}
+
 UCLASS_DRIVER(rtc) = {
 	.name		= "rtc",
 	.id		= UCLASS_RTC,
 	.flags		= DM_UC_FLAG_SEQ_ALIAS,
-#if CONFIG_IS_ENABLED(OF_REAL)
-	.post_bind	= dm_scan_fdt_dev,
-#endif
+	.post_bind	= rtc_post_bind,
 };
diff --git a/include/dm/uclass-id.h b/include/dm/uclass-id.h
index 3ba69ad9a0..50877890b3 100644
--- a/include/dm/uclass-id.h
+++ b/include/dm/uclass-id.h
@@ -83,6 +83,7 @@ enum uclass_id {
 	UCLASS_NOP,		/* No-op devices */
 	UCLASS_NORTHBRIDGE,	/* Intel Northbridge / SDRAM controller */
 	UCLASS_NVME,		/* NVM Express device */
+	UCLASS_NVMEM,		/* Non-volatile memory devices */
 	UCLASS_P2SB,		/* (x86) Primary-to-Sideband Bus */
 	UCLASS_PANEL,		/* Display panel, such as an LCD */
 	UCLASS_PANEL_BACKLIGHT,	/* Backlight controller for panel */
diff --git a/include/nvmem.h b/include/nvmem.h
index 822e698bdd..230841499d 100644
--- a/include/nvmem.h
+++ b/include/nvmem.h
@@ -6,18 +6,8 @@
 #ifndef NVMEM_H
 #define NVMEM_H
 
-/**
- * DOC: Design
- *
- * The NVMEM subsystem is a "meta-uclass" in that it abstracts over several
- * different uclasses all with read/write APIs. One approach to implementing
- * this could be to add a new sub-device for each nvmem-style device of
- * UCLASS_NVMEM. This subsystem has taken the approach of using the existing
- * access methods (i2c_eeprom_write, misc_write, etc.) directly. This has the
- * advantage of not requiring an extra device/driver, saving on binary size and
- * runtime memory usage. On the other hand, it is not idiomatic. Similar
- * efforts should generally use a new uclass.
- */
+struct driver;
+struct udevice;
 
 /**
  * struct nvmem_cell - One datum within non-volatile memory
@@ -31,8 +21,6 @@ struct nvmem_cell {
 	size_t size;
 };
 
-struct udevice;
-
 #if CONFIG_IS_ENABLED(NVMEM)
 
 /**
@@ -104,6 +92,22 @@ int nvmem_cell_get_by_index(struct udevice *dev, int index,
 int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
 			   struct nvmem_cell *cell);
 
+/**
+ * nvmem_register() - Register a new nvmem device
+ * @dev: The nvmem device proper
+ * @drv: The nvmem adapter driver
+ *
+ * This registers a child device of @dev as an nvmem device using @drv. The
+ * child device's name will be @dev's name with ".nvmem" appended. @drv should
+ * contain %DM_FLAG_NAME_ALLOCED in its flags.
+ *
+ * Return:
+ * * 0 on success
+ * * -ENOMEM if no memory could be allocated for the name
+ * * A negative error if device_bind() failed
+ */
+int nvmem_register(struct udevice *dev, const struct driver* drv);
+
 #else /* CONFIG_NVMEM */
 
 static inline int nvmem_cell_read(struct nvmem_cell *cell, void *buf, int size)
@@ -129,6 +133,53 @@ static inline int nvmem_cell_get_by_name(struct udevice *dev, const char *name,
 	return -ENOSYS;
 }
 
+static inline int nvmem_register(struct udevice *dev, const struct driver* drv)
+{
+	return 0;
+}
 #endif /* CONFIG_NVMEM */
 
+/**
+ * struct nvmem_ops - Ops implemented by nvmem devices
+ * read: Read a register
+ * write: Write a register
+ */
+struct nvmem_ops {
+	int (*read)(struct udevice *dev, unsigned int offset, void *buf,
+		    size_t size);
+	int (*write)(struct udevice *dev, unsigned int offset, const void *buf,
+		     size_t size);
+};
+
+#if 0 /* for docs only */
+/**
+ * read() - Read a register from an nvmem device
+ *
+ * @dev: The device to read from
+ * @offset: The offset of the register from the beginning of @dev
+ * @buf: The buffer to read into
+ * @size: The size of @buf, in bytes
+ *
+ * Return:
+ * * 0 on success
+ * * A negative error on failure
+ */
+int read(struct udevice *dev, unsigned int offset, void *buf, size_t size);
+
+/**
+ * write() - Write a register to an nvmem device
+ * @dev: The device to write
+ * @offset: The offset of the register from the beginning of @dev
+ * @buf: The buffer to write
+ * @size: The size of @buf, in bytes
+ *
+ * Return:
+ * * 0 on success
+ * * -ENOSYS if the device is read-only
+ * * A negative error on other failures
+ */
+int write(struct udevice *dev, unsigned int offset, const void *buf,
+	  size_t size);
+#endif
+
 #endif /* NVMEM_H */
-- 
2.35.1.1320.gc452695387.dirty


^ permalink raw reply related	[flat|nested] 19+ messages in thread

* Re: [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe
  2022-05-05 17:11 ` [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe Sean Anderson
@ 2022-05-18 14:29   ` Vladimir Oltean
  0 siblings, 0 replies; 19+ messages in thread
From: Vladimir Oltean @ 2022-05-18 14:29 UTC (permalink / raw)
  To: Sean Anderson
  Cc: u-boot, Simon Glass, Heinrich Schuchardt, Mario Six, Ramon Fried,
	Tom Rini, Joe Hershberger, Alexandru Marginean, Claudiu Manoil

On Thu, May 05, 2022 at 01:11:36PM -0400, Sean Anderson wrote:
> If the DSA master fails to probe for whatever reason, then DSA devices
> will continue on as if nothing is wrong. This can cause incorrect
> behavior. In particular, on sandbox, dsa_sandbox_probe attempts to
> access the master's private data. This is only safe to do if the master
> has been probed first. Fix this by probing the master after we look it
> up, and bailing out if we get an error.
> 
> Fixes: fc054d563b ("net: Introduce DSA class for Ethernet switches")
> Signed-off-by: Sean Anderson <sean.anderson@seco.com>
> ---

Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>

^ permalink raw reply	[flat|nested] 19+ messages in thread

* Re: [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices
  2022-05-05 17:11 ` [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices Sean Anderson
@ 2022-06-09 19:21   ` Tom Rini
  0 siblings, 0 replies; 19+ messages in thread
From: Tom Rini @ 2022-06-09 19:21 UTC (permalink / raw)
  To: Sean Anderson
  Cc: u-boot, Simon Glass, Heinrich Schuchardt, Mario Six, Ramon Fried,
	Joe Hershberger, Marek Behún

[-- Attachment #1: Type: text/plain, Size: 1238 bytes --]

On Thu, May 05, 2022 at 01:11:30PM -0400, Sean Anderson wrote:

> Commit f3dd213e15 ("net: introduce helpers to get PHY ofnode from MAC")
> changed the ethernet sequence assignment from
> 
> uclass 36: ethernet
> 0   * eth@10002000 @ 05813460, seq 0
> 1   * eth@10003000 @ 05813550, seq 5
> 2   * sbe5 @ 05813640, seq 3
> 3   * eth@10004000 @ 05813730, seq 6
> 4   * dsa-test-eth @ 05813820, seq 4
> 5   * lan0 @ 05813a30, seq 2
> 6   * lan1 @ 05813b50, seq 7
> 
> to
> 
> uclass 36: ethernet
> 0   * eth@10002000 @ 03813630, seq 0
> 1   * eth@10003000 @ 03813720, seq 5
> 2   * sbe5 @ 03813810, seq 3
> 3   * eth@10004000 @ 03813900, seq 6
> 4     phy-test-eth @ 038139f0, seq 7
> 5   * dsa-test-eth @ 03813ae0, seq 4
> 6   * lan0 @ 03813cf0, seq 2
> 7   * lan1 @ 03813e10, seq 8
> 
> This caused the mac address assignment to switch around. Avoid this in
> the future by assigning aliases for all ethernet devices. This reverts
> the sequence to what it was before the aformentioned commit (with
> phy-test-eth as seq 8). There is no ethernet1 for whatever reason.
> 
> Signed-off-by: Sean Anderson <sean.anderson@seco.com>

For the series and patches 1-15, applied to u-boot/next, thanks!

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

^ permalink raw reply	[flat|nested] 19+ messages in thread

end of thread, other threads:[~2022-06-09 19:21 UTC | newest]

Thread overview: 19+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-05-05 17:11 [PATCH v4 00/16] Add support for NVMEM API Sean Anderson
2022-05-05 17:11 ` [PATCH v4 01/16] sandbox: net: Add aliases for ethernet devices Sean Anderson
2022-06-09 19:21   ` Tom Rini
2022-05-05 17:11 ` [PATCH v4 02/16] sandbox: net: Add mac address for eth8 to environment Sean Anderson
2022-05-05 17:11 ` [PATCH v4 03/16] test: eth: Add test for ethernet addresses Sean Anderson
2022-05-05 17:11 ` [PATCH v4 04/16] sandbox: net: Remove fake-host-hwaddr Sean Anderson
2022-05-05 17:11 ` [PATCH v4 05/16] sandbox: Remove eth2addr from environment Sean Anderson
2022-05-05 17:11 ` [PATCH v4 06/16] sandbox: Move some mac addresses to device tree Sean Anderson
2022-05-05 17:11 ` [PATCH v4 07/16] net: dsa: Fix segmentation fault if master fails to probe Sean Anderson
2022-05-18 14:29   ` Vladimir Oltean
2022-05-05 17:11 ` [PATCH v4 08/16] misc: i2c_eeprom: Make i2c_eeprom_write use a const buf Sean Anderson
2022-05-05 17:11 ` [PATCH v4 09/16] misc: i2c_eeprom: Add fallbacks Sean Anderson
2022-05-05 17:11 ` [PATCH v4 10/16] misc: Add support for nvmem cells Sean Anderson
2022-05-05 17:11 ` [PATCH v4 11/16] sandbox: Enable NVMEM Sean Anderson
2022-05-05 17:11 ` [PATCH v4 12/16] net: Add support for reading mac addresses from nvmem cells Sean Anderson
2022-05-05 17:11 ` [PATCH v4 13/16] test: Load mac address with i2c eeprom Sean Anderson
2022-05-05 17:11 ` [PATCH v4 14/16] test: Load mac address using RTC Sean Anderson
2022-05-05 17:11 ` [PATCH v4 15/16] test: Load mac address using misc device Sean Anderson
2022-05-05 17:11 ` [PATCH v4 16/16] [RFC] misc: nvmem: Convert to using udevices Sean Anderson

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.