[PATCH] ARM: cpuidle: Avoid memleak if init fail

[Leo Yan]

On Fri, Sep 01, 2017 at 09:08:59AM +0200, Stefan Wahren wrote:
> Hi Leo,
> 
> Am 01.09.2017 um 08:56 schrieb Leo Yan:
> >On Fri, Sep 01, 2017 at 07:49:29AM +0200, Stefan Wahren wrote:
> >>Hi Leo,
> >>
> >>>Leo Yan <leo.yan at linaro.org> hat am 1. September 2017 um 03:33 geschrieben:
> >>>
> >>>
> >>>Hi Stefan,
> >>>
> >>>On Thu, Aug 31, 2017 at 10:24:36PM +0200, Stefan Wahren wrote:
> >>>>In case there are no DT idle states defined or
> >>>>cpuidle_register_driver() fails, the copy of the idle driver is leaked:
> >>>>
> >>>>     unreferenced object 0xede0dc00 (size 1024):
> >>>>     comm "swapper/0", pid 1, jiffies 4294937431 (age 744.510s)
> >>>>     hex dump (first 32 bytes):
> >>>>     94 9e 0b c1 00 00 00 00 00 00 00 00 00 00 00 00 ................
> >>>>     57 46 49 00 00 00 00 00 00 00 00 00 00 00 00 00 WFI.............
> >>>>     backtrace:
> >>>>     [<c1295f04>] arm_idle_init+0x44/0x1ac
> >>>>     [<c0301e6c>] do_one_initcall+0x3c/0x16c
> >>>>     [<c1200d70>] kernel_init_freeable+0x110/0x1d0
> >>>>     [<c0cb3624>] kernel_init+0x8/0x114
> >>>>     [<c0307a98>] ret_from_fork+0x14/0x3c
> >>>>
> >>>>So fix this by freeing the unregistered copy in error case.
> >>>>
> >>>>Signed-off-by: Stefan Wahren <stefan.wahren at i2se.com>
> >>>>Fixes: d50a7d8acd78 ("ARM: cpuidle: Support asymmetric idle definition")
> >>>>---
> >>>>  drivers/cpuidle/cpuidle-arm.c | 6 ++++--
> >>>>  1 file changed, 4 insertions(+), 2 deletions(-)
> >>>>
> >>>>diff --git a/drivers/cpuidle/cpuidle-arm.c b/drivers/cpuidle/cpuidle-arm.c
> >>>>index 7080c38..52a7505 100644
> >>>>--- a/drivers/cpuidle/cpuidle-arm.c
> >>>>+++ b/drivers/cpuidle/cpuidle-arm.c
> >>>>@@ -104,13 +104,13 @@ static int __init arm_idle_init(void)
> >>>>  		ret = dt_init_idle_driver(drv, arm_idle_state_match, 1);
> >>>>  		if (ret <= 0) {
> >>>>  			ret = ret ? : -ENODEV;
> >>>>-			goto out_fail;
> >>>>+			goto init_fail;
> >>>>  		}
> >>>>  		ret = cpuidle_register_driver(drv);
> >>>>  		if (ret) {
> >>>>  			pr_err("Failed to register cpuidle driver\n");
> >>>>-			goto out_fail;
> >>>>+			goto init_fail;
> >>>>  		}
> >>>>  		/*
> >>>>@@ -149,6 +149,8 @@ static int __init arm_idle_init(void)
> >>>>  	}
> >>>>  	return 0;
> >>>>+init_fail:
> >>>>+	kfree(drv);
> >>>The below loop only releases resource for previous CPUs, so should check
> >>>two variables 'drv' and 'dev'. If 'dev != NULL', we also need to release
> >>>it.
> >>i cannot see a leak for 'dev', because this is already handled in the error case of cpuidle_register_device before jumping to out_fail. I agree this isn't consistent, but this is a fix which should go to stable. So only necessary changes.
> >You are right, so please ignore my comment.
> >
> >BTW, to avoid missing anything, are you working on the second
> >fixing to correct the previous CPU resourcee releasing? Actually
> >I think you are welcome to work on the second fixing, but if
> >not I will take the second fixing with rebasing your patch.
> 
> i'm busy with other issues and the RPi doesn't have DT idle states
> so i would be grateful if you could take care of this.

Got it, sure, will work out the patch for reviewing.

Thanks,
Leo Yan