On 02.09.2025 00:27, [email protected] wrote:
> On Mon, Sep 01, 2025 at 10:14:04AM +0200, Jan Beulich wrote:
>> On 29.08.2025 21:27, Stefano Stabellini wrote:
>>> On Thu, 28 Aug 2025, [email protected] wrote:
>>>> --- /dev/null
>>>> +++ b/xen/common/emul/vuart/vuart.c
>>>> @@ -0,0 +1,156 @@
>>>> +/* SPDX-License-Identifier: GPL-2.0-only */
>>>> +/*
>>>> + * UART emulator framework.
>>>> + *
>>>> + * Copyright 2025 Ford Motor Company
>>>> + */
>>>> +
>>>> +#include <xen/err.h>
>>>> +#include <xen/sched.h>
>>>> +#include <xen/vuart.h>
>>>> +#include <xen/xvmalloc.h>
>>>> +
>>>> +#define for_each_emulator(e) \
>>>> +    for ( e = vuart_array_start; e < vuart_array_end; e++ )
>>>> +
>>>> +extern const struct vuart_emulator vuart_array_start[];
>>>> +extern const struct vuart_emulator vuart_array_end[];
>>>> +
>>>> +static const struct vuart_emulator *
>>>> +vuart_match_by_compatible(struct domain *d, const char *compat)
>>>> +{
>>>> +    const struct vuart_emulator *emulator;
>>>> +
>>>> +    if ( d->console.vuart )
>>>> +        return NULL;
>>>> +
>>>> +    for_each_emulator(emulator)
>>>> +        if ( emulator->compatible &&
>>>> +             !strncmp(emulator->compatible, compat,
>>>> +                      strlen(emulator->compatible)) )
>>>
>>> strncmp will continue until the given count even if compat is shorter
>>
>> Not really, one string having a nul char and the other not having one is a
>> difference, at which point comparison will stop. There would be a problem
>> if "compat" didn't point to a nul-terminated string, though (and I didn't
>> check that aspect, not the least because then "shorter" doesn't really
>> make much sense without a length passed in).
> 
> re: NUL-termination: current assumption is that both compat and
> emulator->compatible are NUL-terminated.
> 
> Current `compat` comes from the hardcoded NUL-terminated string (vuart_info).
> 
> In case of `compat` is not NUL-terminated (I plan to populate the field from
> xl in the future), strncmp() will stop after strlen(emulator->compatible)
> bytes.

Which might be too late; if not nul-terminated, "compat" may extend past a
page boundary, with the latter page not mapped.

Jan

Reply via email to