As for the sexism bit, I think the reaction would probably have been different had the submitter (edit: poster, rather: David Howells submitted the code, but it was mjg that the "explosion" was directed at) merely been just a "Red Hat Engineer" (which mjg is not anymore, IIRC). Garrett is a core guy, and Linus knows him well and respects his work. He chose that language to make a broader point, not merely to "explode". This isn't about emotions, it's about content.
Which is sort of the point here: the "shim signed by Microsoft" trick isn't the "Wrong Thing", really. The hardware companies absolutely need a top-level cert authority if secure boot is going to have any utility at all, and Microsoft is the only entity that has stepped up to the plate. What else are we going to do? Dell and Lenovo and Acer aren't about to add a key to their database (and test it) just because the Linux Foundation asked nicely. They're shipping windows hardware. That's not going to happen. So the Shim idea is as good as we're going to get, and as Red Hat (and SUSE, and Canonical) has to ship a product that runs on real hardware, it's the choice they had to make.
That said, I think there's a strong argument that baking this authority into the kernel is a bad idea. Linus clearly agrees (and frankly mjg probably does too, but you can't just write a feature and not try to upstream it). And he's made that point as forcefully and publicly as he can.
Microsoft is the OS vendor. They aren't the only stakeholder in this space. There was, for a time, hope that the UEFI Forum, or an entity attached to it, or perhaps Intel itself might step up and offer to be a central key authority for the UEFI secure boot standard. They didn't. Microsoft did, and even offered to sign keys for other OSes.
Which is sort of the point here: the "shim signed by Microsoft" trick isn't the "Wrong Thing", really. The hardware companies absolutely need a top-level cert authority if secure boot is going to have any utility at all, and Microsoft is the only entity that has stepped up to the plate. What else are we going to do? Dell and Lenovo and Acer aren't about to add a key to their database (and test it) just because the Linux Foundation asked nicely. They're shipping windows hardware. That's not going to happen. So the Shim idea is as good as we're going to get, and as Red Hat (and SUSE, and Canonical) has to ship a product that runs on real hardware, it's the choice they had to make.
That said, I think there's a strong argument that baking this authority into the kernel is a bad idea. Linus clearly agrees (and frankly mjg probably does too, but you can't just write a feature and not try to upstream it). And he's made that point as forcefully and publicly as he can.