-
Notifications
You must be signed in to change notification settings - Fork 4.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove System.IO.Hashing dependency from Nrbf #103700
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is not acceptable, please close the PR.
#endif | ||
return (int)XxHash32.HashToUInt32(MemoryMarshal.AsBytes(integers)); | ||
} | ||
public override int GetHashCode() => _id; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is done on purpose, to prevent from hashcode collision attacks (the attacker can create a payload with such Ids that would cause OOM in the dictionary)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There's no randomization in xxhash32. this is just a deterministic mapping from one integer to another, with the only possibility of creating more collisions. to mitigate such attacks, this should use a random seed, eg System.HashCode.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There's no randomization in xxhash32. this is just a deterministic mapping from one integer to another, with the only possibility of creating more collisions. to mitigate such attacks, this should use a random seed, eg System.HashCode
Thanks, I was not aware of that. I've assumed that it's the right thing to do since it got approved by @bartonjs.
This library needs to target Full Framework and .NET Standard, so we can not use System.HashCode
for all TFMs. But... w can use it for #if NET
. Would that solve the problem for windowsdesktop repo?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
so we can not use System.HashCode for all TFMs
We can:
https://www.nuget.org/packages/Microsoft.Bcl.HashCode/
But if the goal here is just to include some randomness with a single integer, I'd assume we can just do something like:
private static readonly int s_seed = new Random().Next(-int.MaxValue, int.MaxValue);
public int GetHashCode() => _id ^ s_seed;
or some other form of incorporation that Jeremy and Levi bless.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We can: https://www.nuget.org/packages/Microsoft.Bcl.HashCode/
Nice, I was not aware of the fact that such package exists. It's already used by Cbor, so we should be fine using it?
<PackageReference Include="Microsoft.Bcl.HashCode" Version="$(MicrosoftBclHashCodeVersion)" /> |
But if the goal here is just to include some randomness with a single integer
or some other form of incorporation that Jeremy and Levi bless.
I would prefer to wait for @bartonjs or @GrabYourPitchforks feedback.
@stephentoub would you mind updating your PR to switch to HashCode
type or would you prefer me to do it?
@ViktorHofer, does this need any addition to NetCoreAppLibrary.props? Presumably no because the Microsoft.Bcl.HashCode dependency is only downlevel? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, thank you for raising the question, sharing the knowledge and providing the fix!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice. This again needs a reaction in windowsdesktop.
Presumably no because the Microsoft.Bcl.HashCode dependency is only downlevel?
Exactly. This change is a net positive for net9.0 (and only that flows to windowsdesktop) as it removes the IO.Hashing dependency without introducing a new one.
Is this a typo? I'm confused as based on the second part of your response I thought I didn't need to do anything else in windowsdesktop. |
Nope, no typo. This PR needs to be updated again to remove the System.IO.Hashing entry: dotnet/windowsdesktop#4459 Reason is, the WindowsDesktop shared framework needs to be closure complete (so it must not have any dangling references) which requires listing all of the assemblies + its references in the sfxprojs. Now that we remove that dependency again, we need to remove that line from the sfxprojs. |
Oh, I thought you were saying this PR (as in 103700) needs further updates. But you're not saying that; you're saying a subsequent change needs to be made in dotnet/windowsdesktop, right? |
Exactly. This PR is ready to go. Sorry for the confusion. |
Cool, thanks. |
/ba-g failures are known and unrelated |
#103695 (comment)