On 14 November 2013 03:14, Rafael J. Wysocki <rjw@xxxxxxx> wrote: > On Wednesday, November 13, 2013 01:35:38 PM Amit Pundir wrote: >> On 13 November 2013 05:29, Rafael J. Wysocki <rjw@xxxxxxx> wrote: >> > On Wednesday, November 13, 2013 02:22:28 AM Amit Pundir wrote: >> >> ep_create_wakeup_source() reports ENOMEM >> > >> > That needs to be fixed too. I suppose we can make the wakeup_source_register() >> > stub for CONFIG_PM_SLEEP unset return ERR_PTR(-ENOSYS) or something like that >> > and ep_create_wakeup_source() return that instead of -ENOMEM. It looks like >> > eventpoll.c is the only user of it built for CONFIG_PM_SLEEP unset, but that >> > needs to be double checked. >> >> Instead of modifying wakeup_source_register() stub, what if I make >> ep_create_wakeup_source() static inline as well and use its stub to >> return -ENOSYS when CONFIG_PM_SLEEP is not set? >> ep_create_wakeup_source() is used only in fs/eventpoll.c anyway. > > Well, you can do that too. > On second thought we may skip modifying ep_create_wakeup_source() or wakeup_source_register() altogether because once we drop EPOLLWAKEUP from epoll events mask(if PM_SLEEP is unset) then I don't see us running into ep_create_wakeup_source() again. And the only reason for ep_create_wakeup_source() failure will be -ENOMEM as far as I can see. >> >> if wakeup_source_register() >> >> returns NULL. ep_create_wakeup_source() assumes that NULL is only >> >> returned if we run into ENOMEM but NULL is also returned when >> >> CONFIG_PM_SLEEP is disabled. >> >> >> >> Signed-off-by: Amit Pundir <amit.pundir@xxxxxxxxxx> >> >> --- >> >> Changed in v2: >> >> Using static inline functions instead of #ifdefs >> >> --- >> >> fs/eventpoll.c | 3 +-- >> >> include/uapi/linux/eventpoll.h | 12 ++++++++++++ >> >> 2 files changed, 13 insertions(+), 2 deletions(-) >> >> >> >> diff --git a/fs/eventpoll.c b/fs/eventpoll.c >> >> index 473e09d..10f9c43 100644 >> >> --- a/fs/eventpoll.c >> >> +++ b/fs/eventpoll.c >> >> @@ -1820,8 +1820,7 @@ SYSCALL_DEFINE4(epoll_ctl, int, epfd, int, op, int, fd, >> >> goto error_tgt_fput; >> >> >> >> /* Check if EPOLLWAKEUP is allowed */ >> >> - if ((epds.events & EPOLLWAKEUP) && !capable(CAP_BLOCK_SUSPEND)) >> >> - epds.events &= ~EPOLLWAKEUP; >> >> + ep_epollwakeup_check(&epds.events); >> > >> > The "check" part of the name kind of suggests that the function will not change >> > things. What about ep_adjust_epollwakeup() or something along these lines? >> >> I see couple of ep_set_* functions in eventpoll.c. Does it make sense >> to have something like ep_set_epollwakeup()? > > This particular one doesn't really set anything. I suppose that a name like > "ep_take_care_of_epollwakeup" might be somewhat closer to what it really does ... I'm running out of ideas on this one, lets go with "ep_take_care_of_epollwakeup". Regards, Amit Pundir > > Thanks! > > -- > I speak only for myself. > Rafael J. Wysocki, Intel Open Source Technology Center. -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
