Before the reboot, /proc/buddyinfo would have possibly explained this case.
On Monday, 11 December 2017 10:59:18 UTC-8, Tony Guan wrote: > > Update: upon a reboot, I managed to reserve the huge pages with one shoot > so this is not a blocker for me now. > > But still, that 'no whipping, no go' is weird... > > On Monday, December 11, 2017 at 10:42:43 AM UTC-8, Tony Guan wrote: >> >> First off: uname -a >> >> Linux xguan-wsl 3.13.0-112-generic #159-Ubuntu SMP Fri Mar 3 15:26:07 UTC >> 2017 x86_64 x86_64 x86_64 GNU/Linux >> >> As the ZGC (cr.openjdk.java.net/~pliden/zgc/) is opensourced, I was >> curious enough to build it and run it for the first time in my life. >> >> Yet before I can make it work, I need to config the page pool to have the >> required number pages (say, 9216 pages), then the weird thing happens: >> >> It seems that the nr_hugepages is incremented little by little whenever I >> touch it: >> >> The first time I run this command: >> echo 9216 > /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages; cat >> /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages >> >> I got a 20 back. >> >> Then second time I run this: >> 32 >> >> But when I run cat /sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages >> alone, that number will be stuck to 32 (as a cross-checking, "cat >> /proc/sys/vm/nr_hugepages" gives out the same number too). >> >> I understand the kernel is trying hard to accommodate to the big page >> number that I wanted, and I may not get the desired huge page number I want >> immediately, but this behavior seems to be weird in that you have to write >> to it in order for the kernel to do some work. >> >> Just want to check if anybody else is observing the same thing? >> >> Thanks. >> >> >> -- You received this message because you are subscribed to the Google Groups "mechanical-sympathy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
