Re: [Crash-utility] [RFC PATCH 1/2] crash: Show memory overcommit data in dump_kmeminfo()

Aaron, I like the addition. But I've got a few nits regarding the patch. Given that "kmem -i" is so commonly-used, I prefer not to introduce something that could conceivably generate a command-killing "invalid structure member offset" error. And that could happen in your hugetlb_total_pages() function because of the hstate structure dependencies. (See dump_hstates() for example) Also, in dump_kmeminfo() you use MEMBER_OFFSET(), which should be avoided because it will quietly return -1 if the structure member doesn't exist, leading to bogus output. That's the whole purpose behind using OFFSET(), or at least storing/validating the MEMBER_OFFSET() return value before blindly using it.