Re: how to unmap pages in an anonymous mmap?

[Xiubo Li]

On 2017å02æ28æ 03:32, Andy Grover wrote:
> On 02/26/2017 09:59 PM, Xiubo Li wrote:
> > > But, We likely don't want to release memory from the data area anyways
> > > while active, in any case. How about if we set a timer when active
> > > commands go to zero, and then reduce data area to some minimum if no new
> > > cmds come in before timer expires?
> > If I understand correctly: for example, we have 1G(as the minimum)
> > data area and all blocks have been allocated and mapped to runner's
> > vma, then we extern it to 1G + 256M as needed. When there have no
> > active cmds and after the timer expires, will it reduce the data area
> > back to 1G ? And then should it release the reduced 256M data area's
> > memories ?
> >
> > If so, after kfree()ed the blocks' memories, it should also try to remove
> > all the ptes which are mapping this page(like using the try_to_umap()),
> > but something like try_to_umap() doesn't export for the modules.
> >
> > Without ummaping the kfree()ed pages' ptes mentioned above, then
> > the reduced 256M vma space couldn't be reused again for the runner
> > process, because the runner has already do the mapping for the reduced
> > vma space to some old physical pages(won't trigger new page fault
> > again). Then there will be a hole, and the hole will be bigger and bigger.
> >
> > Without ummaping the kfree()ed pages' ptes mentioned above, the
> > pages' reference count (page_ref_dec(), which _inc()ed in page fault)
> > couldn't be reduced back too.
> Let's ask people who will know...
>
> Hi linux-mm,
>
> TCM-User (drivers/target/target_core_user.c) currently uses vmalloc()ed
> memory to back a ring buffer that is mmaped by userspace.
>
> We want to move to dynamically mapping pages into this region, and also
> we'd like to unmap/free pages when idle. What's the right way to unmap?
> I see unmap_mapping_range() but that mentions an underlying file, which
> TCMU doesn't have. Or maybe zap_page_range()? But it's not exported.
Hi linux-mm

For the TCMU case, the vm is not anonymous mapping. And still has
device file desc:

mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_SHARED, dev->fd, 0);

If using the unmap_mapping_range() to do the dynamically maping,
is it okay ? Any other potential risks ?

Or the mentioned 'underlying file' is must one desk file ?

Thanks very much,

BRs
Xiubo



> Any advice?
>
> Thanks in advance -- Regards -- Andy