[PATCH nft] meta time: use uint64_t instead of time_t

[PATCH nft] meta time: use uint64_t instead of time_t

[Lukas Straub]

[-- Attachment #1: Type: text/plain, Size: 836 bytes --]

time_t may be 32 bit on some platforms and thus can't fit a timestamp
with nanoseconds resolution. This causes overflows and ultimatively
breaks meta time expressions on such platforms.

Fix this by using uint64_t instead.

Closes: https://bugzilla.netfilter.org/show_bug.cgi?id=1567
Fixes: f8f32deda31df597614d9f1f64ffb0c0320f4d54 
("meta: Introduce new conditions 'time', 'day' and 'hour'")
Signed-off-by: Lukas Straub <lukasstraub2@web.de>

Index: b/src/meta.c
===================================================================
--- a/src/meta.c
+++ b/src/meta.c
@@ -444,7 +444,7 @@ static struct error_record *date_type_pa
 					    struct expr **res)
 {
 	const char *endptr = sym->identifier;
-	time_t tstamp;
+	uint64_t tstamp;
 
 	if ((tstamp = parse_iso_date(sym->identifier)) != -1)
 		goto success;


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH nft] meta time: use uint64_t instead of time_t

[Pablo Neira Ayuso]

On Tue, Apr 05, 2022 at 10:41:14AM +0000, Lukas Straub wrote:
> time_t may be 32 bit on some platforms and thus can't fit a timestamp
> with nanoseconds resolution. This causes overflows and ultimatively
> breaks meta time expressions on such platforms.
> 
> Fix this by using uint64_t instead.

Applied, thanks

Re: [PATCH nft] meta time: use uint64_t instead of time_t

[Phil Sutter]

On Tue, Apr 05, 2022 at 10:41:14AM +0000, Lukas Straub wrote:
> time_t may be 32 bit on some platforms and thus can't fit a timestamp
> with nanoseconds resolution. This causes overflows and ultimatively
> breaks meta time expressions on such platforms.
> 
> Fix this by using uint64_t instead.
> 
> Closes: https://bugzilla.netfilter.org/show_bug.cgi?id=1567
> Fixes: f8f32deda31df597614d9f1f64ffb0c0320f4d54 
> ("meta: Introduce new conditions 'time', 'day' and 'hour'")
> Signed-off-by: Lukas Straub <lukasstraub2@web.de>
> 
> Index: b/src/meta.c
> ===================================================================
> --- a/src/meta.c
> +++ b/src/meta.c
> @@ -444,7 +444,7 @@ static struct error_record *date_type_pa
>  					    struct expr **res)
>  {
>  	const char *endptr = sym->identifier;
> -	time_t tstamp;
> +	uint64_t tstamp;
>  
>  	if ((tstamp = parse_iso_date(sym->identifier)) != -1)

Doesn't this introduce a warning due to signed/unsigned comparison?

I guess you'll have to cast -1 to uint64_t as well.

Cheers, Phil