During the integration of the RTL8239 POE chip + its frontend MCU, it was noticed that multi-byte operations were basically broken in the current driver.
Tests using SMBus Block Writes showed that the data (after the Wr maker + Ack) was mixed up on the wire. At first glance, it looked like an endianness problem. But for transfers where the number of count + data bytes was not divisible by 4, the last bytes were not looking like an endianness problem because they were in the wrong order but not for example 0 - which would be the case for an endianness problem with 32 bit registers. At the end, it turned out to be the way how i2c_write tried to add the bytes to the send registers.
Each 32 bit register was used similar to a shift register - shifting the various bytes up the register while the next one is added to the least significant byte. But the I2C controller expects the first byte of the transmission in the least significant byte of the first register. And the last byte (assuming it is a 16 byte transfer) is expected in the most significant byte of the fourth register.
While doing these tests, it was also observed that the count byte was missing from the SMBus Block Writes. The driver just removed them from the data->block (from the I2C subsystem). But the I2C controller DOES NOT automatically add this byte - for example by using the configured transmission length.
The RTL8239 MCU is not actually an SMBus compliant device. Instead, it expects I2C Block Reads + I2C Block Writes. But according to the already identified bugs in the driver, it was clear that the I2C controller can simply be modified to not send the count byte for I2C_SMBUS_I2C_BLOCK_DATA. The receive part just needs to write the content of the receive buffer to the correct position in data->block.
While the on-wire format was now correct, reads were still not possible against the MCU (for the RTL8239 POE chip). It was always timing out because the 2ms were not enough for sending the read request and then receiving the 12 byte answer.
These changes were originally submitted to OpenWrt. But there are plans to migrate OpenWrt to the upstream Linux driver. As a result, the pull request was stopped and the changes were redone against this driver.
For reasons of transparency: The work on I2C_SMBUS_I2C_BLOCK_DATA support for the RTL8239-MCU was done on RTL931xx. All problems were therefore detected with the patches from Jonas Jelonek [1] and not the vanilla Linux driver. But looking through the code, it seems like these are NOT regressions introduced by the RTL931x patchset.
I've picked up Alex Guo's patch [2] to reduce conflicts between pending fixes.
[1] https://patchwork.ozlabs.org/project/linux-i2c/cover/20250727114800.3046-1-j... [2] https://lore.kernel.org/r/20250615235248.529019-1-alexguo1023@gmail.com
Signed-off-by: Sven Eckelmann sven@narfation.org --- Changes in v5: - Simplify function/capability registration by using I2C_FUNC_SMBUS_I2C_BLOCK, thanks Jonas Jelonek - Link to v4: https://lore.kernel.org/r/20250809-i2c-rtl9300-multi-byte-v4-0-d71dd5eb6121@...
Changes in v4: - Provide only "write" examples for "i2c: rtl9300: Fix multi-byte I2C write" - drop the second initialization of vals in rtl9300_i2c_write() directly in the "Fix multi-byte I2C write" fix - indicate in target branch for each patch in PATCH prefix - minor commit message cleanups - Link to v3: https://lore.kernel.org/r/20250804-i2c-rtl9300-multi-byte-v3-0-e20607e1b28c@...
Changes in v3: - integrated patch https://lore.kernel.org/r/20250615235248.529019-1-alexguo1023@gmail.com to avoid conflicts in the I2C_SMBUS_BLOCK_DATA code - added Fixes and stable@vger.kernel.org to Alex Guo's patch - added Chris Packham's Reviewed-by/Acked-by - Link to v2: https://lore.kernel.org/r/20250803-i2c-rtl9300-multi-byte-v2-0-9b7b759fe2b6@...
Changes in v2: - add the missing transfer width and read length increase for the SMBus Write/Read - Link to v1: https://lore.kernel.org/r/20250802-i2c-rtl9300-multi-byte-v1-0-5f687e0098e2@...
--- Alex Guo (1): i2c: rtl9300: Fix out-of-bounds bug in rtl9300_i2c_smbus_xfer
Harshal Gohel (2): [i2c-host-fixes] i2c: rtl9300: Fix multi-byte I2C write [i2c-host] i2c: rtl9300: Implement I2C block read and write
Sven Eckelmann (2): [i2c-host-fixes] i2c: rtl9300: Increase timeout for transfer polling [i2c-host-fixes] i2c: rtl9300: Add missing count byte for SMBus Block Ops
drivers/i2c/busses/i2c-rtl9300.c | 51 +++++++++++++++++++++++++++++++++------- 1 file changed, 42 insertions(+), 9 deletions(-) --- base-commit: 7e161a991ea71e6ec526abc8f40c6852ebe3d946 change-id: 20250802-i2c-rtl9300-multi-byte-edaa1fb0872c
Best regards,
From: Alex Guo alexguo1023@gmail.com
The data->block[0] variable comes from user. Without proper check, the variable may be very large to cause an out-of-bounds bug.
Fix this bug by checking the value of data->block[0] first.
1. commit 39244cc75482 ("i2c: ismt: Fix an out-of-bounds bug in ismt_access()") 2. commit 92fbb6d1296f ("i2c: xgene-slimpro: Fix out-of-bounds bug in xgene_slimpro_i2c_xfer()")
Cc: stable@vger.kernel.org Fixes: c366be720235 ("i2c: Add driver for the RTL9300 I2C controller") Signed-off-by: Alex Guo alexguo1023@gmail.com Reviewed-by: Chris Packham chris.packham@alliedtelesis.co.nz Tested-by: Chris Packham chris.packham@alliedtelesis.co.nz Reviewed-by: Wolfram Sang wsa+renesas@sang-engineering.com Signed-off-by: Sven Eckelmann sven@narfation.org --- drivers/i2c/busses/i2c-rtl9300.c | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/drivers/i2c/busses/i2c-rtl9300.c b/drivers/i2c/busses/i2c-rtl9300.c index e064e8a4a1f0824abc82fa677866b85f99fbe4a7..568495720810b373c4fa3b31d3f4cdec7c64b5f9 100644 --- a/drivers/i2c/busses/i2c-rtl9300.c +++ b/drivers/i2c/busses/i2c-rtl9300.c @@ -281,6 +281,10 @@ static int rtl9300_i2c_smbus_xfer(struct i2c_adapter *adap, u16 addr, unsigned s ret = rtl9300_i2c_reg_addr_set(i2c, command, 1); if (ret) goto out_unlock; + if (data->block[0] < 1 || data->block[0] > I2C_SMBUS_BLOCK_MAX) { + ret = -EINVAL; + goto out_unlock; + } ret = rtl9300_i2c_config_xfer(i2c, chan, addr, data->block[0]); if (ret) goto out_unlock;
From: Harshal Gohel hg@simonwunderlich.de
The RTL93xx I2C controller has 4 32 bit registers to store the bytes for the upcoming I2C transmission. The first byte is stored in the least-significant byte of the first register. And the last byte in the most significant byte of the last register. A map of the transferred bytes to their order in the registers is:
reg 0: 0x04_03_02_01 reg 1: 0x08_07_06_05 reg 2: 0x0c_0b_0a_09 reg 3: 0x10_0f_0e_0d
The i2c_read() function basically demonstrates how the hardware would pick up bytes from this register set. But the i2c_write() function was just pushing bytes one after another to the least significant byte of a register AFTER shifting the last one to the next more significant byte position.
If you would then have tried to send a buffer with numbers 1-11 using i2c_write(), you would have ended up with following register content:
reg 0: 0x01_02_03_04 reg 1: 0x05_06_07_08 reg 2: 0x00_09_0a_0b reg 3: 0x00_00_00_00
On the wire, you would then have seen:
Sr Addr Wr [A] 04 A 03 A 02 A 01 A 08 A 07 A 06 A 05 A 0b A 0a A 09 A P
But the correct data transmission was expected to be
Sr Addr Wr [A] 01 A 02 A 03 A 04 A 05 A 06 A 07 A 08 A 09 A 0a A 0b A P
Because of this multi-byte ordering problem, only single byte i2c_write() operations were executed correctly (on the wire).
By shifting the byte directly to the correct end position in the register, it is possible to avoid this incorrect byte ordering and fix multi-byte transmissions.
The second initialization (to 0) of vals was also be dropped because this array is initialized to 0 on the stack by using `= {};`. This makes the fix a lot more readable.
Cc: stable@vger.kernel.org Fixes: c366be720235 ("i2c: Add driver for the RTL9300 I2C controller") Signed-off-by: Harshal Gohel hg@simonwunderlich.de Co-developed-by: Sven Eckelmann sven@narfation.org Signed-off-by: Sven Eckelmann sven@narfation.org Reviewed-by: Chris Packham chris.packham@alliedtelesis.co.nz Tested-by: Chris Packham chris.packham@alliedtelesis.co.nz --- drivers/i2c/busses/i2c-rtl9300.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/i2c/busses/i2c-rtl9300.c b/drivers/i2c/busses/i2c-rtl9300.c index 568495720810b373c4fa3b31d3f4cdec7c64b5f9..4a538b2660802c98e1a51d2ed782e154f2a5d1a0 100644 --- a/drivers/i2c/busses/i2c-rtl9300.c +++ b/drivers/i2c/busses/i2c-rtl9300.c @@ -143,10 +143,10 @@ static int rtl9300_i2c_write(struct rtl9300_i2c *i2c, u8 *buf, int len) return -EIO;
for (i = 0; i < len; i++) { - if (i % 4 == 0) - vals[i/4] = 0; - vals[i/4] <<= 8; - vals[i/4] |= buf[i]; + unsigned int shift = (i % 4) * 8; + unsigned int reg = i / 4; + + vals[reg] |= buf[i] << shift; }
return regmap_bulk_write(i2c->regmap, i2c->reg_base + RTL9300_I2C_MST_DATA_WORD0,
The timeout for transfers was only set to 2ms. Because of this relatively low limit, 12-byte read operations to the frontend MCU of a RTL8239 POE PSE chip cluster was consistently resulting in a timeout.
The original OpenWrt downstream driver [1] was not using any timeout limit at all. This is also possible by setting the timeout_us parameter of regmap_read_poll_timeout() to 0. But since the driver currently implements the ETIMEDOUT error, it is more sensible to increase the timeout in such a way that communication with the (quite common) Realtek I2C-connected POE management solution is possible.
[1] https://git.openwrt.org/?p=openwrt/openwrt.git%3Ba=blob%3Bf=target/linux/rea...
Cc: stable@vger.kernel.org Fixes: c366be720235 ("i2c: Add driver for the RTL9300 I2C controller") Signed-off-by: Sven Eckelmann sven@narfation.org Reviewed-by: Chris Packham chris.packham@alliedtelesis.co.nz Tested-by: Chris Packham chris.packham@alliedtelesis.co.nz --- drivers/i2c/busses/i2c-rtl9300.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/i2c/busses/i2c-rtl9300.c b/drivers/i2c/busses/i2c-rtl9300.c index 4a538b2660802c98e1a51d2ed782e154f2a5d1a0..4a282d57e2c1a72c95bdabdd9eb348a73df28c44 100644 --- a/drivers/i2c/busses/i2c-rtl9300.c +++ b/drivers/i2c/busses/i2c-rtl9300.c @@ -175,7 +175,7 @@ static int rtl9300_i2c_execute_xfer(struct rtl9300_i2c *i2c, char read_write, return ret;
ret = regmap_read_poll_timeout(i2c->regmap, i2c->reg_base + RTL9300_I2C_MST_CTRL1, - val, !(val & RTL9300_I2C_MST_CTRL1_I2C_TRIG), 100, 2000); + val, !(val & RTL9300_I2C_MST_CTRL1_I2C_TRIG), 100, 100000); if (ret) return ret;
The expected on-wire format of an SMBus Block Write is
S Addr Wr [A] Comm [A] Count [A] Data [A] Data [A] ... [A] Data [A] P
Everything starting from the Count byte is provided by the I2C subsystem in the array data->block. But the driver was skipping the Count byte (data->block[0]) when sending it to the RTL93xx I2C controller.
Only the actual data could be seen on the wire:
S Addr Wr [A] Comm [A] Data [A] Data [A] ... [A] Data [A] P
This wire format is not SMBus Block Write compatible but matches the format of an I2C Block Write. Simply adding the count byte to the buffer for the I2C controller is enough to fix the transmission.
This also affects read because the I2C controller must receive the count byte + $count * data bytes.
Cc: stable@vger.kernel.org Fixes: c366be720235 ("i2c: Add driver for the RTL9300 I2C controller") Signed-off-by: Sven Eckelmann sven@narfation.org Reviewed-by: Chris Packham chris.packham@alliedtelesis.co.nz Tested-by: Chris Packham chris.packham@alliedtelesis.co.nz --- drivers/i2c/busses/i2c-rtl9300.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/i2c/busses/i2c-rtl9300.c b/drivers/i2c/busses/i2c-rtl9300.c index 4a282d57e2c1a72c95bdabdd9eb348a73df28c44..cfafe089102aa208dde37096d5105d4140278ca9 100644 --- a/drivers/i2c/busses/i2c-rtl9300.c +++ b/drivers/i2c/busses/i2c-rtl9300.c @@ -285,15 +285,15 @@ static int rtl9300_i2c_smbus_xfer(struct i2c_adapter *adap, u16 addr, unsigned s ret = -EINVAL; goto out_unlock; } - ret = rtl9300_i2c_config_xfer(i2c, chan, addr, data->block[0]); + ret = rtl9300_i2c_config_xfer(i2c, chan, addr, data->block[0] + 1); if (ret) goto out_unlock; if (read_write == I2C_SMBUS_WRITE) { - ret = rtl9300_i2c_write(i2c, &data->block[1], data->block[0]); + ret = rtl9300_i2c_write(i2c, &data->block[0], data->block[0] + 1); if (ret) goto out_unlock; } - len = data->block[0]; + len = data->block[0] + 1; break;
default:
Hi Sven,
Alex Guo (1): i2c: rtl9300: Fix out-of-bounds bug in rtl9300_i2c_smbus_xfer
Harshal Gohel (2): [i2c-host-fixes] i2c: rtl9300: Fix multi-byte I2C write [i2c-host] i2c: rtl9300: Implement I2C block read and write
Sven Eckelmann (2): [i2c-host-fixes] i2c: rtl9300: Increase timeout for transfer polling [i2c-host-fixes] i2c: rtl9300: Add missing count byte for SMBus Block Ops
All merged into i2c/i2c-host-fixes except for patch 5 which has been merged intou i2c/i2c-host.
Thanks everyone for the reviews and tests.
Andi
linux-stable-mirror@lists.linaro.org
-
Andi Shyti -
Sven Eckelmann