Updates to amd_sfh
Subsystems (2)
Basavaraj Natikar
maintainercover Updates to amd_sfh
This patch series include changes for renaming the float32 variable including the fix for shift-out-of-bounds issue.
Basavaraj Natikar (2):
HID: amd_sfh: Rename the float32 variable
HID: amd_sfh: Fix for shift-out-of-bounds
drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c | 30 ++++++++++++++----- 1 file changed, 23 insertions(+), 7 deletions(-)
Signature · 1 line
2.25.1
Basavaraj Natikar
maintainerPATCH 1/2 HID: amd_sfh: Rename the float32 variable
As float32 is also used in other places as a data type, it is necessary to rename the float32 variable in order to avoid confusion.
--- drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c +6 -6
@@ -132,13 +132,13 @@ static void get_common_inputs(struct common_input_property *common, int report_i common->event_type =HID_USAGE_SENSOR_EVENT_DATA_UPDATED_ENUM ; } -static int float_to_int (u32 float32) +static int float_to_int (u32 flt32_val) {int fraction, shift, mantissa, sign, exp, zeropre; - mantissa = float32 &GENMASK (22 ,0 ); - sign = (float32 &BIT (31 )) ?-1 :1 ; - exp = (float32 & ~BIT (31 )) >>23 ; + mantissa = flt32_val &GENMASK (22 ,0 ); + sign = (flt32_val &BIT (31 )) ?-1 :1 ; + exp = (flt32_val & ~BIT (31 )) >>23 ;if (!exp && !mantissa)return 0 ;
@@ -151,10 +151,10 @@ static int float_to_int(u32 float32) } shift =23 - exp; - float32 =BIT (exp) + (mantissa >> shift); + flt32_val =BIT (exp) + (mantissa >> shift); fraction = mantissa &GENMASK (shift -1 ,0 ); -return (((fraction *100 ) >> shift) >=50 ) ? sign * (float32 +1 ) : sign * float32; +return (((fraction *100 ) >> shift) >=50 ) ? sign * (flt32_val +1 ) : sign * flt32_val; } static u8 get_input_rep(u8 current_index, int sensor_idx, int report_id,
Signature · 1 line
2.25.1
Basavaraj Natikar
maintainerPATCH 2/2 HID: amd_sfh: Fix for shift-out-of-bounds
In reply to Basavaraj NatikarShift operation of 'exp' and 'shift' variables exceeds the maximum number of shift values in the u32 range leading to UBSAN shift-out-of-bounds.
...
[ 6.120512] UBSAN: shift-out-of-bounds in drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c:149:50 [ 6.120598] shift exponent 104 is too large for 64-bit type 'long unsigned int' [ 6.120659] CPU: 4 PID: 96 Comm: kworker/4:1 Not tainted 6.4.0amd_1-next-20230519-dirty #10 [ 6.120665] Hardware name: AMD Birman-PHX/Birman-PHX, BIOS SFH_with_HPD_SEN.FD 04/05/2023 [ 6.120667] Workqueue: events amd_sfh_work_buffer [amd_sfh] [ 6.120687] Call Trace: [ 6.120690] <TASK> [ 6.120694] dump_stack_lvl+0x48/0x70 [ 6.120704] dump_stack+0x10/0x20 [ 6.120707] ubsan_epilogue+0x9/0x40 [ 6.120716] __ubsan_handle_shift_out_of_bounds+0x10f/0x170 [ 6.120720] ? psi_group_change+0x25f/0x4b0 [ 6.120729] float_to_int.cold+0x18/0xba [amd_sfh] [ 6.120739] get_input_rep+0x57/0x340 [amd_sfh] [ 6.120748] ? __schedule+0xba7/0x1b60 [ 6.120756] ? __pfx_get_input_rep+0x10/0x10 [amd_sfh] [ 6.120764] amd_sfh_work_buffer+0x91/0x180 [amd_sfh] [ 6.120772] process_one_work+0x229/0x430 [ 6.120780] worker_thread+0x4a/0x3c0 [ 6.120784] ? __pfx_worker_thread+0x10/0x10 [ 6.120788] kthread+0xf7/0x130 [ 6.120792] ? __pfx_kthread+0x10/0x10 [ 6.120795] ret_from_fork+0x29/0x50 [ 6.120804] </TASK>
...
Fix this by adding the condition to validate shift ranges.
--- drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c | 20 +++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-)
drivers/hid/amd-sfh-hid/sfh1_1/amd_sfh_desc.c +18 -2
@@ -143,16 +143,32 @@ static int float_to_int(u32 flt32_val)if (!exp && !mantissa)return 0 ; +/* +* Calculate the exponent and fraction part of floating +* point representation. +*/ exp -=127 ;if (exp <0 ) { exp = -exp; +if (exp >=BITS_PER_TYPE (u32)) +return 0 ; zeropre = (((BIT (23 ) + mantissa) *100 ) >>23 ) >> exp;return zeropre >=50 ? sign :0 ; } shift =23 - exp; - flt32_val =BIT (exp) + (mantissa >> shift); - fraction = mantissa &GENMASK (shift -1 ,0 ); +if (abs (shift) >=BITS_PER_TYPE (u32)) +return 0 ; + +if (shift <0 ) { + shift = -shift; + flt32_val =BIT (exp) + (mantissa << shift); + shift =0 ; + }else { + flt32_val =BIT (exp) + (mantissa >> shift); + } + + fraction = (shift ==0 ) ?0 : mantissa &GENMASK (shift -1 ,0 );return (((fraction *100 ) >> shift) >=50 ) ? sign * (flt32_val +1 ) : sign * flt32_val; }
Signature · 1 line
2.25.1
Quoted 8 lines · Basavaraj Natikar
This patch series include changes for renaming the float32 variable including the fix for shift-out-of-bounds issue.
Basavaraj Natikar (2):
HID:amd_sfh: Rename the float32 variable
HID:amd_sfh: Fix for shift-out-of-bounds[...]
Applied to https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git (for-6.5/upstream-fixes), thanks!
[1/2] HID: amd_sfh: Rename the float32 variable
https://git.kernel.org/hid/hid/c/c1685a862a4b [2/2] HID: amd_sfh: Fix for shift-out-of-bounds https://git.kernel.org/hid/hid/c/878543661764
Cheers,
Signature · 1 line
Benjamin Tissoires <[email protected]>