linux-kernel - RE: [PATCH] drivers core: Free dma_range

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <SJ0PR11MB5072D1748184147EA9534709F1D10@SJ0PR11MB5072.namprd11.prod.outlook.com>
Date:   Tue, 5 Jan 2021 02:46:57 +0000
From:   "Li, Meng" <Meng.Li@...driver.com>
To:     Jim Quinlan <jim2101024@...il.com>,
        Jim Quinlan <james.quinlan@...adcom.com>,
        open list <linux-kernel@...r.kernel.org>
CC:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "rafael@...nel.org" <rafael@...nel.org>,
        "Hao, Kexin" <Kexin.Hao@...driver.com>,
        Christoph Hellwig <hch@....de>
Subject: RE: [PATCH] drivers core: Free dma_range_map when driver probe failed



> -----Original Message-----
> From: Jim Quinlan <jim2101024@...il.com>
> Sent: Monday, January 4, 2021 11:56 PM
> To: Jim Quinlan <james.quinlan@...adcom.com>; Li, Meng
> <Meng.Li@...driver.com>; open list <linux-kernel@...r.kernel.org>
> Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>; rafael@...nel.org;
> Hao, Kexin <Kexin.Hao@...driver.com>; Jim Quinlan
> <jim2101024@...il.com>; Christoph Hellwig <hch@....de>
> Subject: Re: [PATCH] drivers core: Free dma_range_map when driver probe
> failed
> 
> > Subject: [PATCH] drivers core: Free dma_range_map when driver probe
> > failed
> >
> > There will be memory leak if driver probe failed. Trace as below:
> >   backtrace:
> >     [<000000002415258f>] kmemleak_alloc+0x3c/0x50
> >     [<00000000f447ebe4>] __kmalloc+0x208/0x530
> >     [<0000000048bc7b3a>] of_dma_get_range+0xe4/0x1b0
> >     [<0000000041e39065>] of_dma_configure_id+0x58/0x27c
> >     [<000000006356866a>] platform_dma_configure+0x2c/0x40
> >     ......
> >     [<000000000afcf9b5>] ret_from_fork+0x10/0x3c
> >
> > This issue is introduced by commit e0d072782c73("dma-mapping:
> > introduce DMA range map, supplanting dma_pfn_offset "). It doesn't
> > free dma_range_map when driver probe failed and cause above memory
> > leak. So, add code to free it in error path.
> >
> > Fixes: e0d072782c73("dma-mapping: introduce DMA range map,
> supplanting
> > dma_pfn_offset ")
> > Cc: stable@...r.kernel.org
> > Signed-off-by: Meng Li <Meng.Li@...driver.com>
> > ---
> >  drivers/base/dd.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/base/dd.c b/drivers/base/dd.c index
> > 148e81969e04..8e4871aa34bc 100644
> > --- a/drivers/base/dd.c
> > +++ b/drivers/base/dd.c
> > @@ -612,6 +612,7 @@ static int really_probe(struct device *dev, struct
> > device_driver *drv)
> >         else if (drv->remove)
> >                 drv->remove(dev);
> >  probe_failed:
> > +       kfree(dev->dma_range_map);
> >         if (dev->bus)
> >                 blocking_notifier_call_chain(&dev->bus->p->bus_notifier,
> >
> > BUS_NOTIFY_DRIVER_NOT_BOUND, dev);
> > --
> > 2.17.1
> >
> > --
> 
> Hi Meng,
> 
> Sorry for the delay -- I was on vacation.  I agree with you -- thanks.
> However, note that in function  device_release()  in drivers/base/core.c
> there is this line:
> 
> kfree(dev->dma_range_map);
> 
> Won't this also be called  if all of the appropriate drivers' probes fail for this
> device, effecting a double kfree? 

If there are more than one driver compatible with the device, all the drivers will allocate memory for dma_range_map in function dev->bus->dma_configure().
If all the drivers probe failed, they all need to free memory. So, I think there is no double kfree case.

 Perhaps your patch could also set "dev-
> >dma_range_map" to NULL after calling kfree()?

I agree with you. Based on all drivers probe failed case, it is need to set dma_range_map  as NULL after calling kfree() to avoid the NULL pointer.
In additional, could you please help to check whether we also need to set dma_range_map  as NULL in device_release()?
Because I found out some other cases(for example, all drivers register failed) may free dma_range_map  multiple times.

Thanks,
Limeng


> 
> Thanks much,
> Jim Quinlan
> Broadcom STB