View Single Post
  #342 (permalink)  
Old 04-01-2012, 12:41 PM
arrrghhh's Avatar
arrrghhh
Testing Extraordinaire
Offline
Threadstarter
 
Join Date: Mar 2007
Posts: 3,604
Reputation: 7360
arrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the communityarrrghhh is a trusted member of the community
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Re: [TESTING] XDAndroid GBX0B 2.3.7 Build

Quote:
Originally Posted by auburnham View Post
I had an opportunity to investigate this further. I compared all files changed from the 9/20/11 rootfs.img to the 9/30/11 rootfs.img, and identified the change made by mweirauch labeled "fixup /dev/block/ramzswap0 creation" to be in /bin/userinit.sh, line 317. Originally, it was the following:
Code:
mknod -m 0666 /dev/block/ramzswap0 b 254 0
And it was updated to be the following:
Code:
mknod /dev/block/ramzswap0 b 253 0
When I revert back to the original line, (i.e. ... -m 0666 ... 254 0) my free memory is consistently higher and my performance is better without any obvious side effects. Can anyone explain what this line is doing, and why I might see the effects I'm seeing? Also, just to be clear, I am not recommending that anyone make this change, I am just trying to investigate potential performance improvements.
Hrm, I'll see if emwe remembers why he made that change. The only difference I see is the permissions are not being set on the updated line (no -m 0666) and the file size is 1 byte smaller? b 253 instead of 254.

The manpage on mknod will help you understand a bit more what's going on perhaps?

mknod(1): make block/char special files - Linux man page

As to what the default is for permissions, or why you would be seeing different performance is strange indeed. Have you taken my newest rootfs from this bundle and changed that line? It works, you get tangibly better performance?
Reply With Quote