linux-kernel - Re: [PATCH] staging: rtl8712: Fix memory leak in r8712_init_recv

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] [day] [month] [year] [list]

Message-ID: <CAD-N9QX+5aeugkPDVmZkFUG-Oup3pXWV2uvOOYK1WfJCKnt6Zg@mail.gmail.com>
Date:   Fri, 21 May 2021 21:42:32 +0800
From:   慕冬亮 <mudongliangabcd@...il.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     Larry.Finger@...inger.net, florian.c.schilhabel@...glemail.com,
        rkovhaev@...il.com, straube.linux@...il.com,
        linux-staging@...ts.linux.dev,
        linux-kernel <linux-kernel@...r.kernel.org>,
        syzbot+1c46f3771695bccbdb3a@...kaller.appspotmail.com
Subject: Re: [PATCH] staging: rtl8712: Fix memory leak in r8712_init_recv_priv

On Fri, May 21, 2021 at 9:16 PM Greg KH <gregkh@...uxfoundation.org> wrote:
>
> On Fri, May 21, 2021 at 08:24:58PM +0800, 慕冬亮 wrote:
> > On Fri, May 21, 2021 at 8:18 PM Greg KH <gregkh@...uxfoundation.org> wrote:
> > >
> > > On Fri, May 21, 2021 at 08:08:11PM +0800, Dongliang Mu wrote:
> > > > r871xu_dev_remove failed to call r8712_free_drv_sw() and free the
> > > > resource (e.g., struct urb) due to the failure of firmware loading.
> > > >
> > > > Fix this by invoking r8712_free_drv_sw at the failure site.
> > > >
> > > > Reported-by: syzbot+1c46f3771695bccbdb3a@...kaller.appspotmail.com
> > > > Fixes: b4383c971bc5 ("staging: rtl8712: handle firmware load failure")
> > > > Signed-off-by: Dongliang Mu <mudongliangabcd@...il.com>
> > > > ---
> > > >  drivers/staging/rtl8712/usb_intf.c | 13 ++++++++++---
> > > >  1 file changed, 10 insertions(+), 3 deletions(-)
> > > >
> > > > diff --git a/drivers/staging/rtl8712/usb_intf.c b/drivers/staging/rtl8712/usb_intf.c
> > > > index dc21e7743349..a5190b4250ce 100644
> > > > --- a/drivers/staging/rtl8712/usb_intf.c
> > > > +++ b/drivers/staging/rtl8712/usb_intf.c
> > > > @@ -589,7 +589,7 @@ static int r871xu_drv_init(struct usb_interface *pusb_intf,
> > > >   */
> > > >  static void r871xu_dev_remove(struct usb_interface *pusb_intf)
> > > >  {
> > > > -     struct net_device *pnetdev = usb_get_intfdata(pusb_intf);
> > > > +     struct net_device *pnetdev, *newpnetdev = usb_get_intfdata(pusb_intf);
> > > >       struct usb_device *udev = interface_to_usbdev(pusb_intf);
> > > >
> > > >       if (pnetdev) {
> > >
> > > Did you test this?
> >
> > For now, I only tested this patch in my local workspace. The memory
> > leak does not occur any more.
> >
> > I have pushed a patch testing onto the syzbot dashboard [1]. Now it is
> > in the pending state.
> >
> > [1] https://syzkaller.appspot.com/bug?id=3a325b8389fc41c1bc94de0f4ac437ed13cce584
> >
> > >
> > > I think you just broke the code right here :(
> >
> > If I broke any code logic, I am sorry. However, this patch only adds
> > some code to deallocate some resources when failing to load firmware.
> >
> > Do you mean that I replace pnetdev with the variable - newpnetdev?
>
> Yes, and then the first thing the code does is check the value of
> pnetdev which is totally undefined :(

You are right. Apology for the previous patch. I test my old patch
below in the local workspace, it works.

------------------------------------------------------------------------------------------------------------------------
--- a/drivers/staging/rtl8712/usb_intf.c
+++ b/drivers/staging/rtl8712/usb_intf.c
@@ -597,9 +597,9 @@ static void r871xu_dev_remove(struct usb_interface
*pusb_intf)

  /* never exit with a firmware callback pending */
  wait_for_completion(&padapter->rtl8712_fw_ready);
- pnetdev = usb_get_intfdata(pusb_intf);
+ struct net_device *newpnetdev = usb_get_intfdata(pusb_intf);
  usb_set_intfdata(pusb_intf, NULL);
- if (!pnetdev)
+ if (!newpnetdev)
  goto firmware_load_fail;
  release_firmware(padapter->fw);
  if (drvpriv.drv_registered)
@@ -625,6 +625,14 @@ static void r871xu_dev_remove(struct
usb_interface *pusb_intf)
  */
  if (udev->state != USB_STATE_NOTATTACHED)
  usb_reset_device(udev);
+ if (pnetdev) {
+ struct _adapter *padapter = netdev_priv(pnetdev);
+ /* Stop driver mlme relation timer */
+ //r8712_stop_drv_timers(padapter);
+ //r871x_dev_unload(padapter);
+ r8712_free_drv_sw(padapter);
+ /* udev is already freed in failed fireware loading */
+ }
 }
------------------------------------------------------------------------------------------------------------------------

However, the compiler complains the declaration of newpnetdev. So I
moved the declaration to the beginning, but I forget to initialize
both two variables. :(

I will revise this problem and test it in my local workspace. If it
works, I will resend a v2 patch.

BTW, should I uncomment "r8712_stop_drv_timers" and
"r871x_dev_unload"? I am not very sure about its functionability.

>