Ken Raeburn

I just ran a test on s390x running RHEL7 ... it looks like most of the hash algorithms fail their verification checks on a big-endian system, even some of those...

A lot of these changes look like they're not specific to ppc64le, but they're conditionalized as if they are. Some look questionable -- for example, if you pass CityMurmur a...

wiz0u writes: > If you mean "take the low-order 64-bits from the 128-bits output": > What mathematical proof do I have that these 64-bits are as much > collision-proof as...

I think the isPending flag ought to prevent that, but it's possible there's a bug in that somewhere. I'll take a closer look soon…

I'm not so sure that requestState provides sufficient protection there. It will guard against both timeoutIndexOperations and finishIndexOperation trying to make the same request structure usable again, but timeoutIndexOperations is...

This problem came up because of a race condition between two paths for handling deduplication advice and incorrect handling of a data structure on one of those paths. (One path...

Ah, I mis-"spoke" ... the part of the rewrite with the bug fix isn't in yet, even internally; it's still in the works. Porting the fix applied to older versions...

The fix for the 6.2 code should be getting pushed to github within the next day or so, actually. The rewrite going into "unstable" will be a while longer in...

Given the decoupling of the logical address range and the physical storage size that VDO provides, they’re two separate questions. Shrinking the data LV isn’t hard, in fact: [root@test-host]# df...

It’s not clear to me what you’re trying to accomplish by using a single unified VDO storage device but separate encryption keys per volume. I expect this is probably for...