linux-kernel - Re: [PATCH] vt: add support for smput/rmput escape codes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <7f55fd6b-91ed-453c-902a-0f7c74e201f5@grenoble-inp.org>
Date: Mon, 25 Aug 2025 13:41:02 +0200
From: Calixte Pernot <calixte.pernot@...noble-inp.org>
To: Jiri Slaby <jirislaby@...nel.org>,
 Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: linux-kernel@...r.kernel.org, linux-serial@...r.kernel.org
Subject: Re: [PATCH] vt: add support for smput/rmput escape codes

Hi, thanks for your response.

On 8/25/25 07:47, Jiri Slaby wrote:
> Hi,
>
> On 24. 08. 25, 16:20, Calixte Pernot wrote:
>> Support "\e[?1049h" and "\e[?1049l" escape codes.
>> This patch allows programs to enter and leave alternate screens.
>> This feature is widely available in graphical terminal emulators and 
>> mostly
>> used by fullscreen terminal-based user interfaces such as text editors.
>> Most editors such as vim and nano assume this escape code in not 
>> supported
>> and will not try to print the escape sequence if TERM=linux.
>> To try out this patch, run `TERM=xterm-256color vim` inside a VT.
>>
>> Signed-off-by: Calixte Pernot <calixte.pernot@...noble-inp.org>
> ...
>> --- a/drivers/tty/vt/vt.c
>> +++ b/drivers/tty/vt/vt.c
> ...
>> @@ -1878,6 +1883,46 @@ static int get_bracketed_paste(struct 
>> tty_struct *tty)
>>       return vc->vc_bracketed_paste;
>>   }
>>   +/* console_lock is held */
>> +static void enter_alt_screen(struct vc_data *vc)
>> +{
>> +    unsigned int size = vc->vc_rows * vc->vc_cols * 2;
>> +
>> +    if (vc->vc_saved_screen != NULL)
>> +        return; /* Already inside an alt-screen */
>
> All this is protected by console_lock, right? 

Yes. The 2 functions I added are called by csi_DEC_hl, itself protected 
by console_lock.

console_lock() is called in do_con_write, and the call stack looks like 
this:

do_con_write -> do_con_trol -> csi_DEC -> csi_DEC_hl -> 
enter/leave_alt_screen

>> +    vc->vc_saved_screen = kzalloc(size, GFP_NOWAIT);
>
> Why GFP_NOWAIT?

To be honest I used the same flags as vc->vc_screenbuf, but after 
"reading the F-ing manual", this seems unappropriate here.

"GFP_KERNEL implies GFP_RECLAIM, which means that direct reclaim may be 
triggered under memory pressure; the calling context must be allowed to 
sleep"

I guess we are allowed to sleep here (please correct me if i'm wrong), 
so GFP_KERNEL must be more appropriate.

>> +    if (vc->vc_saved_screen == NULL)
>> +        return;
>> +    memcpy(vc->vc_saved_screen, (u16 *)vc->vc_origin, size);
>
> kmemdup();
>
>> ed_cols, vc->vc_cols);
>> +    unsigned short *src, *dest;
>
> u16
>
>> +
>> +    if (vc->vc_saved_screen == NULL)
>> +        return; /* Not inside an alt-screen */
>> +    for (int r = 0; r < rows; r++) {
>
> unsigned
>> --- a/include/linux/console_struct.h
>> +++ b/include/linux/console_struct.h
>> @@ -159,6 +159,9 @@ struct vc_data {
>>       struct uni_pagedict *uni_pagedict;
>>       struct uni_pagedict **uni_pagedict_loc; /* [!] Location of 
>> uni_pagedict variable for this console */
>>       u32 **vc_uni_lines;            /* unicode screen content */
>> +    unsigned short    *vc_saved_screen;
>
> u16 *
>
>> +    unsigned int    vc_saved_cols;
>> +    unsigned int    vc_saved_rows;
>>       /* additional information is in vt_kern.h */
>>   };
>
> thanks,

Thanks for the review, I'll rewrite the patch with the suggested changes 
and send it back.