Re: is anyone maintaining hifn(4)?
Taylor R Campbell <[email protected]> Wed, 14 Jul 2021 12:38:54 +0000
| Newsgroups | gmane.os.netbsd.devel.crypto |
|---|---|
| Message-ID | <[email protected]> |
> Date: Tue, 13 Jul 2021 00:25:56 -0700 > From: emily ingalls <[email protected]> >=20 > i recently got a card that uses a Hifn 7955, and figured i'd take a > look at the driver for it while i was setting it up. i almost > immediately found issues that would likely cause hangs or crashes > (maybe even damage to the chip, not sure) on some machines, possibly > unexpectedly slow operation on others, and could have unexpected and > hard to detect side effects elsewhere due to possible clock > instability. Can you be more specific about the issues? Can you say more about your use case? > i have a vested interest in having this driver work properly and > securely in NetBSD, and i'm interested in sending in patches and > possibly doing a more extensive audit/rewrite, but it looks like the > driver is currently disabled, and there's been little work the past > few years that hasn't been pulling changes from openbsd. so... is > anyone maintaining it at all? >=20 > would i be better off going to OpenBSD for this, and then trying to > merge updates here once they're accepted there (despite having > little interest in OpenBSD personally)? I was probably the last person to touch it, and it looks like I merged what I thought (a year ago) were the important updates from OpenBSD as of May 2020: http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/pci/hifn7751.c?rev=3D1.70&c= ontent-type=3Dtext/x-cvsweb-markup&only_with_tag=3DMAIN At the time, I fixed several bugs in hifn(4) and verified that my device produced RNG output and that the crypto worked for /dev/crypto (and, with `openssl speed', confirmed my hypothesis that it is a crypto decelerator rather than crypto accelerator, at least relative to the ~10yo Ivy Bridge CPU I was using to test it). What do you mean by `currently disabled'? It looks like hifn(4) is included in amd64/GENERIC and i386/GENERIC, at least. If you have any patches, I'd be happy to take a look. (Can't test on my device right now but I'll eventually have access to it again.)