Re: [PATCH] at_keyboard: Fix keyboards that report IBM PC AT scan codes

[Renaud Métrich]

Hi,

Testing the proposed patch on my old Asus N53SN in Legacy failed: as soon as at_keyboard is selected, the keys are corrupted and it's impossible to do anything.

Digging into this, it appears that query_mode() returns 2 (so set2 needs to be used), but in fact internally the keycode are the ones expected by set1.

Then I tried always executing the set_scancodes() code querying the driver and it ended up working perfectly: for real set2 was then to be used.

Hence the new proposal below (Vladimir please review this, it may be "too dirty"):

commit 5e8d6963c50cd727bfb5614448f9c80743a59de1 (HEAD -> master)
Author: Renaud Métrich <rmetrich@redhat.com>
Date:   Thu Dec 3 09:43:21 2020 +0100

    Force using set_scancodes() code querying the driver.
    
    This solves issues when query_mode() returns 2 but the driver continues
    to send keycodes suitable for set 1 (seen on a Asus N53SN).
    
    Resolves: rhbz#1897587
    
    Signed-off-by: Renaud Métrich <rmetrich@redhat.com>

diff --git a/grub-core/term/at_keyboard.c b/grub-core/term/at_keyboard.c
index 0df54d5..c026034 100644
--- a/grub-core/term/at_keyboard.c
+++ b/grub-core/term/at_keyboard.c
@@ -172,7 +172,14 @@ set_scancodes (void)
       return;
     }
 
-#if !USE_SCANCODE_SET
+/*
+ * Code removed: querying the driver seems to be more robust than just querying
+ * the mode being returned. This was tested on both:
+ * - a Lenovo T460s (query_mode() returning 2 and code below working)
+ * - a Asus N53SN (query_mode() returning 2 but code below now working: driver
+ *   was expecting set1 keycodes)
+ */
+#if 0 && !USE_SCANCODE_SET

Renaud.

On 11/6/20 2:48 PM, Javier Martinez Canillas wrote:

> GRUB has support for both the IBM PC XT (Set 1) and IBM PC AT (Set 2) key
> scan code mappings. But the driver always hardcodes to Set 1, overwriting
> the mode that was queried in the grub_keyboard_controller_init() function.
>
> This breaks keyboard mapping on machines whose firmwares are not using the
> Set 1. For example, this is the case in Lenovo P71 and HP Elitebook 8770W
> laptops that report key scan codes using Set 2.
>
> Instead of hardcoding to Set 1, use the value that was queried before.
>
> Reported-by: Renaud Métrich <rmetrich@redhat.com>
> Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
> ---
>
>  grub-core/term/at_keyboard.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/grub-core/term/at_keyboard.c b/grub-core/term/at_keyboard.c
> index 597111077bd..ed2ea0904c1 100644
> --- a/grub-core/term/at_keyboard.c
> +++ b/grub-core/term/at_keyboard.c
> @@ -165,7 +165,7 @@ set_scancodes (void)
>      }
>  
>  #if !USE_SCANCODE_SET
> -  ps2_state.current_set = 1;
> +  ps2_state.current_set = grub_keyboard_orig_set;
>    return;
>  #else
>  
>

OpenPGP_0x5D129094FB6E4326.asc
Description: application/pgp-keys

OpenPGP_signature
Description: OpenPGP digital signature