Re: Feedback Request: Bug #63894 (2038 issue) Proposed Change

[Andrew Hamilton]

Thank you, I see now there are some functions I may be able to use
already in the GRUB gnulib, I will investigate that.

Thank you for the quick feedback!

- Andrew

On Sat, Aug 17, 2024 at 9:03 AM Vladimir 'phcoder' Serbinenko
<phcoder@gmail.com> wrote:
>
> No, it's license-incompatible. Please don't use that code.
>
> Le sam. 17 août 2024, 16:44, Andrew Hamilton <adhamilt@gmail.com> a écrit :
>>
>> Hello,
>>
>> I'm thinking to try to address bug 63894 (grub_datetime2unixtime()
>> still has the year 2038 problem):
>> https://savannah.gnu.org/bugs/?63894
>>
>> I confirmed the issue is still partially present in
>> include/grub/datetime.h... the following checks are still present and
>> the issue mentioned local variable limitations are also still an
>> issue:
>>
>> grub_datetime2unixtime (const struct grub_datetime *datetime, grub_int64_t 
>> *nix)
>> {
>>   grub_int32_t ret;
>> ...
>>   if (datetime->year > 2038 || datetime->year < 1901)
>>     return 0;
>> ...
>>
>> Simply changing the locals to 64-bit and removing the 2038 limit does
>> not completely solve the issue, at some point past 2038 the result of
>> this algorithm produces incorrect results.
>>
>> To completely solve the issue, I wanted to use the algorithm
>> implemented by the Linux kernel in mktime64 (time.c around line 449):
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/time/time.c?h=v6.11-rc3
>>
>> Would it be acceptable to base the GRUB implementation on the Linux
>> Kernel algorithm above?
>>
>> Thanks,
>> Andrew
>>
>> _______________________________________________
>> Grub-devel mailing list
>> Grub-devel@gnu.org
>> https://lists.gnu.org/mailman/listinfo/grub-devel
>
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> https://lists.gnu.org/mailman/listinfo/grub-devel