usbhid: Interpret 0 length ff effects as infinite (0xffff) length effects
Subsystems (4)
PATCH 1/1 usbhid: Interpret 0 length ff effects as infinite (0xffff) length effects
Greetings,
I started using my Accuforce V2 sim wheel on Linux. I was getting no response from racing simulators through wine, while native linux test tools worked properly. It appears that many real-world applications will send 0 as the replay length, which was resulting in the behavior I was observing (nothing). The PID document does not explicitly state that 0 length effects should be interpreted as infinite, but it does mention null effects being infinite effects.
This patch will interpret 0 length force feedback effects as 0xffff (infinite) length effects, leaving other values for replay length unchanged.
--- drivers/hid/usbhid/hid-pidff.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/hid/usbhid/hid-pidff.c +1 -1
@@ -301,7 +301,7 @@ static void pidff_set_effect_report(struct pidff_device *pidff, pidff->block_load[PID_EFFECT_BLOCK_INDEX ].value[0 ]; pidff->set_effect_type->value[0 ] = pidff->create_new_effect_type->value[0 ]; - pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length; + pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length ==0 ?0xffff : effect->replay.length; pidff->set_effect[PID_TRIGGER_BUTTON ].value[0 ] = effect->trigger.button; pidff->set_effect[PID_TRIGGER_REPEAT_INT ].value[0 ] = effect->trigger.interval;
Signature · 1 line
2.35.1
Paul Dino Jones kirjoitti 2022-10-02 01:16:
Quoted 1 line
Greetings,
Hello, and thanks for looking into this!
Quoted 7 lines
I started using my Accuforce V2 sim wheel on Linux. I was getting no response from racing simulators through wine, while native linux test tools worked properly. It appears that many real-world applications will send 0 as the replay length, which was resulting in the behavior I was observing (nothing). The PID document does not explicitly state that 0 length effects should be interpreted as infinite, but it does mention null effects being infinite effects.
Actually, it is Wine that is translating 0xFFFF from the application to 0x0000 for the Linux FF API:
https://gitlab.winehq.org/wine/wine/-/blob/master/dlls/winebus.sys/bus_udev.c#L1124
Unfortunately "infinite" duration is not actually specified at all in our API currently. input.h just says that the all durations are in msecs and values above 0x7fff cause unspecified results.
We have three places where the duration is handled:
- ff-memless: Considers 0 as infinite (in ml_get_combo_effect() and calculate_next_time()).
- iforce-ff: Just passes the duration to HW as-is - it is unknown what counts as infinite, if any.
- pidff: Just passes the duration to HW as-is, so using the unspecified-by-API 0xffff results in infinite duration (per USB HID PID spec).
So we probably want to specify some value to work as infinite, likely either 0 or 0xFFFF, and explicitly document that in input.h.
I suspect that ff-memless devices are currently the most popular, and e.g. Wine already assumes 0 is infinite, and I can't think of a reason to have an "actual" 0-duration effect, so I guess 0 would be the most sensible value.
Since iforce is an "ancestor" of HID PID of sorts, it may also support 0xffff = infinite. I'll try to get hold of one to test, though it may take a couple of weeks...
Quoted 25 lines
This patch will interpret 0 length force feedback effects as
0xffff(infinite) length effects, leaving other values for replay length unchanged.Signed-off-by Paul Dino Jones <[email protected]>--- drivers/hid/usbhid/hid-pidff.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/hid/usbhid/hid-pidff.c +1
-1@@ -301,7 +301,7 @@ static void pidff_set_effect_report(struct pidff_device *pidff, pidff->block_load[PID_EFFECT_BLOCK_INDEX ].value[0 ]; pidff->set_effect_type->value[0 ] = pidff->create_new_effect_type->value[0 ]; - pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length; + pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length ==0 ?0xffff : effect->replay.length; pidff->set_effect[PID_TRIGGER_BUTTON].value[0] =
effect->trigger.button;drivers/hid/usbhid/hid-pidff.c +0
-0pidff->set_effect[PID_TRIGGER_REPEAT_INT ].value[0 ] = effect->trigger.interval;
Signature · 1 line
Anssi Hannula
Hello, and thank you for considering this.
Yes, Wine 7 breaks a lot of stuff for me, and I've been using Wine 5.x and 6.x through Proton which isn't ideal when trying to isolate problems.
It seems there is atleast some precedence in other force feedback drivers for using 0 as some sort of indicator for an infinite effect:
https://github.com/gotzl/hid-fanatecff/blob/next/hid-ftecff.c#L724
https://github.com/berarma/new-lg4ff/blob/master/hid-lg4ff.c#L762
We also discussed this issue at this thread:
https://github.com/berarma/ffbtools/issues/26
I've also read some indication that the SimCube wheel works on Linux, so I'd be interested in how that is handling this situation.
---
Paul Dino Jones
Quoted 72 lines
Paul Dino Jones kirjoitti 2022-10-02 01:16:
Quoted 1 line
Greetings,
Hello, and thanks for looking into this!
Quoted 7 lines
I started using my Accuforce V2 sim wheel on Linux. I was getting no response from racing simulators through wine, while native linux test tools worked properly. It appears that many real-world applications will send 0 as the replay length, which was resulting in the behavior I was observing (nothing). The PID document does not explicitly state that 0 length effects should be interpreted as infinite, but it does mention null effects being infinite effects.
Actually, it is Wine that is translating
0xFFFFfrom the application to0x0000for the Linux FF API:
https://gitlab.winehq.org/wine/wine/-/blob/master/dlls/winebus.sys/bus_udev.c#L1124Unfortunately "infinite" duration is not actually specified at all in our API currently.
input.hjust says that the all durations are in msecs and values above0x7fffcause unspecified results.We have three places where the duration is handled:
- ff-memless: Considers 0 as infinite (inml_get_combo_effect()andcalculate_next_time()).
- iforce-ff: Just passes the duration to HW as-is - it is unknown what counts as infinite, if any.
- pidff: Just passes the duration to HW as-is, so using the unspecified-by-API0xffffresults in infinite duration (per USB HID PID spec).So we probably want to specify some value to work as infinite, likely either 0 or
0xFFFF, and explicitly document that ininput.h.
I suspect that ff-memless devices are currently the most popular, and e.g. Wine already assumes 0 is infinite, and I can't think of a reason to have an "actual" 0-duration effect, so I guess 0 would be the most sensible value.Since iforce is an "ancestor" of HID PID of sorts, it may also support
0xffff= infinite.
I'll try to get hold of one to test, though it may take a couple of weeks...Quoted 26 lines
This patch will interpret 0 length force feedback effects as
0xffff(infinite) length effects, leaving other values for replay length unchanged.Signed-off-by Paul Dino Jones <[email protected]>--- drivers/hid/usbhid/hid-pidff.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/hid/usbhid/hid-pidff.c +1
-1@@ -301,7 +301,7 @@ static void pidff_set_effect_report(struct pidff_device *pidff, pidff->block_load[PID_EFFECT_BLOCK_INDEX ].value[0 ]; pidff->set_effect_type->value[0 ] = pidff->create_new_effect_type->value[0 ]; - pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length; + pidff->set_effect[PID_DURATION ].value[0 ] = effect->replay.length ==0 ?0xffff : effect->replay.length; pidff->set_effect[PID_TRIGGER_BUTTON].value[0] =
effect->trigger.button;drivers/hid/usbhid/hid-pidff.c +0
-0pidff->set_effect[PID_TRIGGER_REPEAT_INT ].value[0 ] = effect->trigger.interval;Signature · 1 line
Anssi Hannula