Input: tca8418_keypad - fix potential infinite loop and OOB, access on I2C error
Subsystems (3)
PATCH 1/1 Input: tca8418_keypad - fix potential infinite loop and OOB, access on I2C error
From 187224ee38e19fd3f74bfc08d1908c97398fef82 Mon Sep 17 00:00:00 2001
From: Zhian Liang <[email protected]>
Date: Thu, 23 Jul 2026 00:08:15 +0800
Subject: [PATCH] Input: tca8418_keypad - fix potential infinite loop and OOB access on I2C error
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
If the I2C bus returns 0xFF (e.g., due to a stuck bus or device fault), the original code would treat it as a valid key event, leading to two critical issues:
1. The loop in tca8418_read_keypad() would never terminate because the condition "reg <= 0" is false for 0xFF (255). This stalls the threaded IRQ handler indefinitely.
2. The extracted hardware keycode (127) is used to compute row/col indices that exceed the valid range (rows*cols ≤ 80), causing an out-of-bounds read on "keymap[code]" when reporting the key. Fix both by:
- Recognizing 0xFF as an empty FIFO condition (along with 0x00).
- Validating the keycode before calculating row/col, skipping invalid codes and preventing array overrun.
---
drivers/input/keyboard/tca8418_keypad.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
drivers/input/keyboard/tca8418_keypad.c +0 -0
@@ -171,13 +171,20 @@ static void tca8418_read_keypad(struct tca8418_keypad *keypad_data)
break;
}
- /* Assume that key code 0 signifies empty FIFO */
- if (reg <= 0)
+ /* 0x00 = empty FIFO, 0xFF = likely bus fault */
+ if (reg == 0 || reg == 0xFF)
break;
state = reg & KEY_EVENT_VALUE;
code = reg & KEY_EVENT_CODE;
+ /* validate keycode: must be non-zero and within hardware limits */
+ if (code == 0 || code > TCA8418_MAX_ROWS * TCA8418_MAX_COLS){
+ dev_err(&keypad_data->client->dev,
+ "invalid key code: %d\n", code);
+ continue;
+ }
+
row = code / TCA8418_MAX_COLS;
col = code % TCA8418_MAX_COLS;
Signature · 1 line
2.34.1
Dmitry Torokhov
maintainerHi Liang,
Quoted 6 lines · Liang Zhan
From
187224ee38e19fd3f74bfc08d1908c97398fef82Mon Sep 17 00:00:00 2001
From: Zhian Liang <[email protected]>
Date: Thu, 23 Jul 2026 00:08:15 +0800
Subject: [PATCH] Input:tca8418_keypad- fix potential infinite loop and OOB access on I2C error
MIME-Version: 1.0
Quoted 8 lines
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
If the I2C bus returns0xFF(e.g., due to a stuck bus or device fault), the original code would treat it as a valid key event, leading to two critical issues:
1. The loop intca8418_read_keypad()would never terminate because the condition "reg <= 0" is false for0xFF(255). This stalls the threaded IRQ handler indefinitely.
Not everything that Sashiko generates needs to be taken literally. If transfer glitches I expect I2C core signal this properly.
Quoted 18 lines
2. The extracted hardware keycode (127) is used to compute row/col indices that exceed the valid range (rows*cols ≤ 80), causing an out-of-bounds read on "keymap[code]" when reporting the key. Fix both by:
- Recognizing0xFFas an empty FIFO condition (along with0x00).
- Validating the keycode before calculating row/col, skipping invalid codes and preventing array overrun.Signed-off-by Zhian Liang <[email protected]>---
drivers/input/keyboard/tca8418_keypad.c | 11 +++++++++--1 file changed, 9 insertions(+), 2 deletions(-)drivers/input/keyboard/tca8418_keypad.c +0
-0
drivers/input/keyboard/tca8418_keypad.c static void tca8418_read_keypad(struct tca8418_keypad · line 178
Quoted 10 lines
drivers/input/keyboard/tca8418_keypad.c +2
-2@@ -171,13 +171,20 @@ static void tca8418_read_keypad(struct tca8418_keypad *keypad_data) break; } - /* Assume that key code 0 signifies empty FIFO */ - if (reg <= 0) + /* 0x00 = empty FIFO, 0xFF = likely bus fault */ + if (reg == 0 || reg == 0xFF) break; state = reg & KEY_EVENT_VALUE; code = reg & KEY_EVENT_CODE;
Jet's move the check for empty FIFO here:
if (!code) return;
drivers/input/keyboard/tca8418_keypad.c static void tca8418_read_keypad(struct tca8418_keypad · line 180
Quoted 2 lines
drivers/input/keyboard/tca8418_keypad.c +2
-0+ /* validate keycode: must be non-zero and within hardware limits */ + if (code == 0 || code > TCA8418_MAX_ROWS * TCA8418_MAX_COLS){
This check is not sufficient if keypad is configured to use just part of potential matrix. We need to make sure that row and col is within the rows and cold limits we set up for the keypad.
Thanks.
Signature · 1 line
Dmitry
PATCH v2 1/1 Input: tca8418_keypad - fix potential infinite loop and OOB access on invalid keycodes
In reply to Liang ZhanThe current code checks the raw register value against zero to detect an empty FIFO. However, the TCA8418 datasheet defines key code 0 as the empty-FIFO condition. If the hardware returns garbage data, the loop may never terminate and the extracted keycode can cause an out-of-bounds access on keymap[].
Move the empty-FIFO check to operate on the extracted key code so that the loop terminates correctly. Additionally, validate row/column indices against the configured matrix dimensions before indexing into the keymap array, skipping invalid events.
---
changes in v2:
- Moved empty-FIFO check to operate on extracted key code instead of raw register value.
- Added row/col validation against configured matrix dimensions.
- Dropped explicit 0xFF check as requested.
--- drivers/input/keyboard/tca8418_keypad.c | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-)
drivers/input/keyboard/tca8418_keypad.c +16 -3
@@ -114,6 +114,8 @@ struct tca8418_keypad {struct input_dev *input;unsigned int row_shift; +unsigned int rows; +unsigned int cols; };/*
@@ -171,19 +173,28 @@ static void tca8418_read_keypad(struct tca8418_keypad *keypad_data)break ; } -/* Assume that key code 0 signifies empty FIFO */ -if (reg <=0 ) -break ; state = reg &KEY_EVENT_VALUE ; code = reg &KEY_EVENT_CODE ; +/* Key code 0 signifies empty FIFO */ +if (!code) +break ; + row = code /TCA8418_MAX_COLS ; col = code %TCA8418_MAX_COLS ; row = (col) ? row : row -1 ; col = (col) ? col -1 :TCA8418_MAX_COLS -1 ; +/* Validate against configured matrix size */ +if (row >= keypad_data->rows || col >= keypad_data->cols) { +dev_err (&keypad_data->client->dev, +"invalid key code %d (row %d, col %d)\n" , + code, row, col); +continue ; + } + code =MATRIX_SCAN_CODE (row, col, keypad_data->row_shift);input_event (input,EV_MSC ,MSC_SCAN , code);input_report_key (input, keymap[code], state);
@@ -298,6 +309,8 @@ static int tca8418_keypad_probe(struct i2c_client *client) keypad_data->client = client; keypad_data->row_shift = row_shift; + keypad_data->rows = rows; + keypad_data->cols = cols;/* Read key lock register, if this fails assume device not present */ error =tca8418_read_byte (keypad_data,REG_KEY_LCK_EC , ®);
Signature · 1 line
2.34.1
Dmitry Torokhov
maintainerRe: [PATCH v2] Input: tca8418_keypad - fix potential infinite loop and OOB access on invalid keycodes
Quoted 10 lines · Liang Zhan
The current code checks the raw register value against zero to detect an empty FIFO. However, the TCA8418 datasheet defines key code 0 as the empty-FIFO condition. If the hardware returns garbage data, the loop may never terminate and the extracted keycode can cause an out-of-bounds access on keymap[].
Move the empty-FIFO check to operate on the extracted key code so that the loop terminates correctly. Additionally, validate row/column indices against the configured matrix dimensions before indexing into the keymap array, skipping invalid events.
Unfortunately your mailer line-wrapped the patch so it does not apply.
Thanks.
Signature · 1 line
Dmitry
Re: [PATCH v2] Input: tca8418_keypad - fix potential infinite loop and OOB access on invalid keycodes
Hi Dmitry,
v3 was sent here:
https://lore.kernel.org/linux-input/[email protected]/T/#t
It broke the thread because I forgot --in-reply-to. Sorry for the noise.