Messages in this thread | | | From | Mike Frysinger <> | Date | Thu, 27 May 2010 16:21:04 -0400 | Subject | Re: [PATCH 0/2] blackfin: ptrace mm/sram_list fixes |
| |
On Thu, May 27, 2010 at 15:55, Oleg Nesterov wrote: > On 05/26, Oleg Nesterov wrote: >> On 05/25, Mike Frysinger wrote: >> > we probably should add proper locking to this >> > structure though. >> >> Agreed. I'll try to make the trivial patch tomorrow. I think we >> can just use mm->mmap_sem, is_user_addr_valid() needs this lock >> for find_vma() anyway. > > please see the patches.
awesome ! wasnt expecting someone to do the work for us :).
> UNTESTED! the second one certainly needs the review from someone > who knows what this code does ;)
np. after i look them over, i'll suck them into the Blackfin tree and let them settle for a release for our test suites to shake out problems.
> BTW. Obviously sys_sram_alloc() can create multiple sram_list_struct > nodes with the same ->addr (with or without this patch), I hope this > is fine.
how so ? Blackfin is a nommu arch, so there should be no aliasing issues. each of the individual L1 sub-allocators manage a different address range, and none of those should return an address that is already in use. it could happen that the code calling these funcs calls a free func directly instead of going through the lsl free, but that's a bug in the usage, not this code. -mike
| |