Re: fallible_collections crate fails to build on i686 arch

[Fabio Valentini]

On Thu, Feb 2, 2023 at 10:14 PM  wrote:
> Hello everyone
>
> While trying to fix the build of the `rust-fallible_collections` package
> I found out [1] that the crate does not work properly with 32 bit systems.
> I am not quite sure if my findings are 100% accurate but lets say they
> are. What would be the best option here? Adding "ExcludeArch" to the
> spec file?
>
> Thank you for you answers in advance!

Hi!

I'm not familiar with this crate's code (it might be fine, for all I
know), but the tests themselves look like they're just begging for UB
and crashes :)
I think Josh is more familiar with this kind of stuff, but I think it
*should* be safe to just skip the test that crashes for Fedora builds
of rust-fallible_collections.

Side note: You might not even really need this crate - many APIs for
fallible allocations have recently been added to the Rust standard
library.
For example Vec::try_reserve has been stable since Rust 1.57:
https://doc.rust-lang.org/stable/std/vec/struct.Vec.html#method.try_reserve

So depending on what the crate that depends on these APIs needs, you
might be able to ditch the dependency on fallible_collections
entirely.

Fabio

fallible_collections crate fails to build on i686 arch

Well this is awkward, I unretired the `rust-fallible_collections` 
package because it is a dependency of a dependency
but after checking again I don't really need the first dependency.
It is only needed for benchmarks und I recently learned that I can just 
remove such dependencies. So yeah....

Should I just retire it again? I don't mind maintaining it, but if you 
say that it is no longer needed
because the APIs are getting stabilized in the standard library then it 
wouldn't make sense to keep it.

Hussein

On 2/2/23 22:26, Fabio Valentini wrote:
> On Thu, Feb 2, 2023 at 10:14 PM  wrote:
> > Hello everyone
> >
> > While trying to fix the build of the `rust-fallible_collections` package
> > I found out [1] that the crate does not work properly with 32 bit systems.
> > I am not quite sure if my findings are 100% accurate but lets say they
> > are. What would be the best option here? Adding "ExcludeArch" to the
> > spec file?
> >
> > Thank you for you answers in advance!
>
> Hi!
>
> I'm not familiar with this crate's code (it might be fine, for all I
> know), but the tests themselves look like they're just begging for UB
> and crashes :)
> I think Josh is more familiar with this kind of stuff, but I think it
> *should* be safe to just skip the test that crashes for Fedora builds
> of rust-fallible_collections.
>
> Side note: You might not even really need this crate - many APIs for
> fallible allocations have recently been added to the Rust standard
> library.
> For example Vec::try_reserve has been stable since Rust 1.57:
> https://doc.rust-lang.org/stable/std/vec/struct.Vec.html#method.try_reserve
>
> So depending on what the crate that depends on these APIs needs, you
> might be able to ditch the dependency on fallible_collections
> entirely.
>
> Fabio
> _______________________________________________
> Rust mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]
> Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue

fallible_collections crate fails to build on i686 arch

[Josh Stone]

On 2/2/23 1:26 PM, Fabio Valentini wrote:
> I'm not familiar with this crate's code (it might be fine, for all I
> know), but the tests themselves look like they're just begging for UB
> and crashes :)
> I think Josh is more familiar with this kind of stuff, but I think it
> *should* be safe to just skip the test that crashes for Fedora builds
> of rust-fallible_collections.

Yeah, it's a bogus test input:
https://github.com/vcombey/fallible_collections/issues/35#issuecomment-14145...

Even on 64-bit, that would be UB to create the invalid Vec/slice, but it
seems in practice that goes unnoticed and the impossible alloc fails as
the test intends.

fallible_collections crate fails to build on i686 arch

[blinxen]

So what your saying is that because we create a memory layout for a 
single u8 value
and then set it as the pointer for the Vec which we then trick into 
thinking that it has a lot of memory causes UB?
This would mean that what I wrote into the issue is completely wrong? Is 
that correct?

Thanks Josh for clarifying!

Hussein

On 2/3/23 01:44, Josh Stone wrote:
> On 2/2/23 1:26 PM, Fabio Valentini wrote:
> > I'm not familiar with this crate's code (it might be fine, for all I
> > know), but the tests themselves look like they're just begging for UB
> > and crashes :)
> > I think Josh is more familiar with this kind of stuff, but I think it
> > *should* be safe to just skip the test that crashes for Fedora builds
> > of rust-fallible_collections.
>
> Yeah, it's a bogus test input:
> https://github.com/vcombey/fallible_collections/issues/35#issuecomment-14145...
>
> Even on 64-bit, that would be UB to create the invalid Vec/slice, but it
> seems in practice that goes unnoticed and the impossible alloc fails as
> the test intends.
> _______________________________________________
> Rust mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]
> Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue