V2:
Incorporated Alan Sterns review comments:
Thanks for mentioning the READ_ONCE() semantics; I really had no clue.
Note:
I think it's also correct to NOT use the "irqsave" variants
of spin_lock_irq/spin_unlock_irq in "usbfs_increase_memory_usage()".
I am not sure if it's worth it?
V1:
patch for f_fs.c:
> > > > + atomic64_sub(amount, &ffs->mmap_mem_usage);
Greg KH:
> > > Why not use a real lock instead of trying to do a fake one with
> > > this atomic variable?
Ingo:
> > I just took the code from "drivers/usb/core/devio.c",
> > "usbfs_increase_memory_usage()".
> > ...
> > You are of course right: You can easily use a lock here and this
> > makes the intention of the code a lot clearer I guess.
> >
> > I will modify the patch accordingly.
Alan Stern:
> If you also feel like making a similar change to the code in devio.c,
> it would be welcome.
Ingo Rohloff (1):
USB: usbfs: Use a spinlock instead of atomic accesses to tally used
memory.
drivers/usb/core/devio.c | 32 ++++++++++++++++++++++----------
1 file changed, 22 insertions(+), 10 deletions(-)
--
2.17.1
