1. Patchew
  2. linux
  3. View series

[PATCH v3] misc: sgi-gru: fix use-after-free error in gru_set_context_option, gru_fault and gru_handle_user_call_os

Zheng Wang posted 1 patch 3 years, 5 months ago
There is a newer version of this series
drivers/misc/sgi-gru/grufault.c  | 14 ++++++++++++--
drivers/misc/sgi-gru/grumain.c   | 18 ++++++++++++++----
drivers/misc/sgi-gru/grutables.h |  2 +-
3 files changed, 27 insertions(+), 7 deletions(-)
[PATCH v3] misc: sgi-gru: fix use-after-free error in gru_set_context_option, gru_fault and gru_handle_user_call_os
Posted by Zheng Wang 3 years, 5 months ago 
Gts may be freed in gru_check_chiplet_assignment.
The caller still use it after that, UAF happens.

Fix it by introducing a return value to see if it's in error path or not.
Free the gts in caller if gru_check_chiplet_assignment check failed.

Fixes: 55484c45dbec ("gru: allow users to specify gru chiplet 2")
Reported-by: Zheng Wang <hackerzheng666@gmail.com>
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
---
v3:
- add preempt_enable and use VM_FAULT_NOPAGE as failure code by Yejian

v2:
- commit message changes suggested by Greg

v1: https://lore.kernel.org/lkml/CAJedcCzY72jqgF-pCPtx66vXXwdPn-KMagZnqrxcpWw1NxTLaA@mail.gmail.com/
---
 drivers/misc/sgi-gru/grufault.c  | 14 ++++++++++++--
 drivers/misc/sgi-gru/grumain.c   | 18 ++++++++++++++----
 drivers/misc/sgi-gru/grutables.h |  2 +-
 3 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/drivers/misc/sgi-gru/grufault.c b/drivers/misc/sgi-gru/grufault.c
index d7ef61e602ed..864be8a2cb61 100644
--- a/drivers/misc/sgi-gru/grufault.c
+++ b/drivers/misc/sgi-gru/grufault.c
@@ -656,7 +656,9 @@ int gru_handle_user_call_os(unsigned long cb)
 	if (ucbnum >= gts->ts_cbr_au_count * GRU_CBR_AU_SIZE)
 		goto exit;
 
-	gru_check_context_placement(gts);
+	ret = gru_check_context_placement(gts);
+	if (ret)
+		goto err;
 
 	/*
 	 * CCH may contain stale data if ts_force_cch_reload is set.
@@ -677,6 +679,10 @@ int gru_handle_user_call_os(unsigned long cb)
 exit:
 	gru_unlock_gts(gts);
 	return ret;
+err:
+	gru_unlock_gts(gts);
+	gru_unload_context(gts, 1);
+	return VM_FAULT_NOPAGE;
 }
 
 /*
@@ -874,7 +880,7 @@ int gru_set_context_option(unsigned long arg)
 		} else {
 			gts->ts_user_blade_id = req.val1;
 			gts->ts_user_chiplet_id = req.val0;
-			gru_check_context_placement(gts);
+			ret = gru_check_context_placement(gts);
 		}
 		break;
 	case sco_gseg_owner:
@@ -889,6 +895,10 @@ int gru_set_context_option(unsigned long arg)
 		ret = -EINVAL;
 	}
 	gru_unlock_gts(gts);
+	if (ret) {
+		gru_unload_context(gts, 1);
+		ret = VM_FAULT_NOPAGE;
+	}
 
 	return ret;
 }
diff --git a/drivers/misc/sgi-gru/grumain.c b/drivers/misc/sgi-gru/grumain.c
index 9afda47efbf2..952803790818 100644
--- a/drivers/misc/sgi-gru/grumain.c
+++ b/drivers/misc/sgi-gru/grumain.c
@@ -716,9 +716,10 @@ static int gru_check_chiplet_assignment(struct gru_state *gru,
  * chiplet. Misassignment can occur if the process migrates to a different
  * blade or if the user changes the selected blade/chiplet.
  */
-void gru_check_context_placement(struct gru_thread_state *gts)
+int gru_check_context_placement(struct gru_thread_state *gts)
 {
 	struct gru_state *gru;
+	int ret = 0;
 
 	/*
 	 * If the current task is the context owner, verify that the
@@ -727,14 +728,16 @@ void gru_check_context_placement(struct gru_thread_state *gts)
 	 */
 	gru = gts->ts_gru;
 	if (!gru || gts->ts_tgid_owner != current->tgid)
-		return;
+		return ret;
 
 	if (!gru_check_chiplet_assignment(gru, gts)) {
 		STAT(check_context_unload);
-		gru_unload_context(gts, 1);
+		ret = VM_FAULT_NOPAGE;
 	} else if (gru_retarget_intr(gts)) {
 		STAT(check_context_retarget_intr);
 	}
+
+	return ret;
 }
 
 
@@ -919,6 +922,7 @@ vm_fault_t gru_fault(struct vm_fault *vmf)
 	struct gru_thread_state *gts;
 	unsigned long paddr, vaddr;
 	unsigned long expires;
+	int ret;
 
 	vaddr = vmf->address;
 	gru_dbg(grudev, "vma %p, vaddr 0x%lx (0x%lx)\n",
@@ -934,7 +938,13 @@ vm_fault_t gru_fault(struct vm_fault *vmf)
 	mutex_lock(&gts->ts_ctxlock);
 	preempt_disable();
 
-	gru_check_context_placement(gts);
+	ret = gru_check_context_placement(gts);
+	if (ret) {
+		preempt_enable();
+		mutex_unlock(&gts->ts_ctxlock);
+		gru_unload_context(gts, 1);
+		return ret;
+	}
 
 	if (!gts->ts_gru) {
 		STAT(load_user_context);
diff --git a/drivers/misc/sgi-gru/grutables.h b/drivers/misc/sgi-gru/grutables.h
index 5efc869fe59a..f4a5a787685f 100644
--- a/drivers/misc/sgi-gru/grutables.h
+++ b/drivers/misc/sgi-gru/grutables.h
@@ -632,7 +632,7 @@ extern int gru_user_flush_tlb(unsigned long arg);
 extern int gru_user_unload_context(unsigned long arg);
 extern int gru_get_exception_detail(unsigned long arg);
 extern int gru_set_context_option(unsigned long address);
-extern void gru_check_context_placement(struct gru_thread_state *gts);
+extern int gru_check_context_placement(struct gru_thread_state *gts);
 extern int gru_cpu_fault_map_id(void);
 extern struct vm_area_struct *gru_find_vma(unsigned long vaddr);
 extern void gru_flush_all_tlb(struct gru_state *gru);
-- 
2.25.1
Re: [PATCH v3] misc: sgi-gru: fix use-after-free error in gru_set_context_option, gru_fault and gru_handle_user_call_os
Posted by Zheng Yejian 3 years, 5 months ago 
On Wed, 19 Oct 2022 00:52:59 +0800, Zheng Wang <zyytlz.wz@163.com> wrote:
> Gts may be freed in gru_check_chiplet_assignment.
> The caller still use it after that, UAF happens.
>
> Fix it by introducing a return value to see if it's in error path or not.
> Free the gts in caller if gru_check_chiplet_assignment check failed.
>
> Fixes: 55484c45dbec ("gru: allow users to specify gru chiplet 2")
> Reported-by: Zheng Wang <hackerzheng666@gmail.com>
> Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
> ---
> v3:
> - add preempt_enable and use VM_FAULT_NOPAGE as failure code by Yejian

Sorry, maybe I didn't make it clearly enough in
https://lore.kernel.org/lkml/20221009201418.509417-1-zhengyejian1@huawei.com/

What I suggested was that return value of gru_fault() should be some VM_FAULT_XXX,
because this return value would be specially handled in pagefault flow like:
    __do_fault() {
        ......
        ret = vma->vm_ops->fault(vmf);
        ......
    }

But in other functions that are not in pagefault flow, common error code
like -EINVAL is ok. Meanings of every error code can be found in:
https://man7.org/linux/man-pages/man3/errno.3.html

By the way, these are all personal suggestions. You need Reviewed-by Maintainer :)

-- Zheng Yejian

>
> v2:
> - commit message changes suggested by Greg
>
> v1: https://lore.kernel.org/lkml/CAJedcCzY72jqgF-pCPtx66vXXwdPn-KMagZnqrxcpWw1NxTLaA@mail.gmail.com/
> ---
>  drivers/misc/sgi-gru/grufault.c  | 14 ++++++++++++--
>  drivers/misc/sgi-gru/grumain.c   | 18 ++++++++++++++----
>  drivers/misc/sgi-gru/grutables.h |  2 +-
>  3 files changed, 27 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/misc/sgi-gru/grufault.c b/drivers/misc/sgi-gru/grufault.c
> index d7ef61e602ed..864be8a2cb61 100644
> --- a/drivers/misc/sgi-gru/grufault.c
> +++ b/drivers/misc/sgi-gru/grufault.c
> @@ -656,7 +656,9 @@ int gru_handle_user_call_os(unsigned long cb)
> 	if (ucbnum >= gts->ts_cbr_au_count * GRU_CBR_AU_SIZE)
> 		goto exit;
>
> -	gru_check_context_placement(gts);
> +	ret = gru_check_context_placement(gts);
> +	if (ret)
> +		goto err;
>
> 	/*
> 	 * CCH may contain stale data if ts_force_cch_reload is set.
> @@ -677,6 +679,10 @@ int gru_handle_user_call_os(unsigned long cb)
>  exit:
> 	gru_unlock_gts(gts);
> 	return ret;
> +err:
> +	gru_unlock_gts(gts);
> +	gru_unload_context(gts, 1);
> +	return VM_FAULT_NOPAGE;
>  }
>
>  /*
> @@ -874,7 +880,7 @@ int gru_set_context_option(unsigned long arg)
> 		} else {
> 			gts->ts_user_blade_id = req.val1;
> 			gts->ts_user_chiplet_id = req.val0;
> -			gru_check_context_placement(gts);
> +			ret = gru_check_context_placement(gts);
> 		}
> 		break;
> 	case sco_gseg_owner:
> @@ -889,6 +895,10 @@ int gru_set_context_option(unsigned long arg)
> 		ret = -EINVAL;
> 	}
> 	gru_unlock_gts(gts);
> +	if (ret) {
> +		gru_unload_context(gts, 1);
> +		ret = VM_FAULT_NOPAGE;
> +	}
>
> 	return ret;
>  }
> diff --git a/drivers/misc/sgi-gru/grumain.c b/drivers/misc/sgi-gru/grumain.c
> index 9afda47efbf2..952803790818 100644
> --- a/drivers/misc/sgi-gru/grumain.c
> +++ b/drivers/misc/sgi-gru/grumain.c
> @@ -716,9 +716,10 @@ static int gru_check_chiplet_assignment(struct gru_state *gru,
>   * chiplet. Misassignment can occur if the process migrates to a different
>   * blade or if the user changes the selected blade/chiplet.
>   */
> -void gru_check_context_placement(struct gru_thread_state *gts)
> +int gru_check_context_placement(struct gru_thread_state *gts)
>  {
> 	struct gru_state *gru;
> +	int ret = 0;
>
> 	/*
> 	 * If the current task is the context owner, verify that the
> @@ -727,14 +728,16 @@ void gru_check_context_placement(struct gru_thread_state *gts)
> 	 */
> 	gru = gts->ts_gru;
> 	if (!gru || gts->ts_tgid_owner != current->tgid)
> -		return;
> +		return ret;
>
> 	if (!gru_check_chiplet_assignment(gru, gts)) {
> 		STAT(check_context_unload);
> -		gru_unload_context(gts, 1);
> +		ret = VM_FAULT_NOPAGE;
> 	} else if (gru_retarget_intr(gts)) {
> 		STAT(check_context_retarget_intr);
> 	}
> +
> +	return ret;
>  }
>
>
> @@ -919,6 +922,7 @@ vm_fault_t gru_fault(struct vm_fault *vmf)
> 	struct gru_thread_state *gts;
> 	unsigned long paddr, vaddr;
> 	unsigned long expires;
> +	int ret;
>
> 	vaddr = vmf->address;
> 	gru_dbg(grudev, "vma %p, vaddr 0x%lx (0x%lx)\n",
> @@ -934,7 +938,13 @@ vm_fault_t gru_fault(struct vm_fault *vmf)
> 	mutex_lock(&gts->ts_ctxlock);
> 	preempt_disable();
>
> -	gru_check_context_placement(gts);
> +	ret = gru_check_context_placement(gts);
> +	if (ret) {
> +		preempt_enable();
> +		mutex_unlock(&gts->ts_ctxlock);
> +		gru_unload_context(gts, 1);
> +		return ret;
> +	}
>
> 	if (!gts->ts_gru) {
> 		STAT(load_user_context);
> diff --git a/drivers/misc/sgi-gru/grutables.h b/drivers/misc/sgi-gru/grutables.h
> index 5efc869fe59a..f4a5a787685f 100644
> --- a/drivers/misc/sgi-gru/grutables.h
> +++ b/drivers/misc/sgi-gru/grutables.h
> @@ -632,7 +632,7 @@ extern int gru_user_flush_tlb(unsigned long arg);
>  extern int gru_user_unload_context(unsigned long arg);
>  extern int gru_get_exception_detail(unsigned long arg);
>  extern int gru_set_context_option(unsigned long address);
> -extern void gru_check_context_placement(struct gru_thread_state *gts);
> +extern int gru_check_context_placement(struct gru_thread_state *gts);
>  extern int gru_cpu_fault_map_id(void);
>  extern struct vm_area_struct *gru_find_vma(unsigned long vaddr);
>  extern void gru_flush_all_tlb(struct gru_state *gru);
> --
> 2.25.1
Re: [PATCH v3] misc: sgi-gru: fix use-after-free error in gru_set_context_option, gru_fault and gru_handle_user_call_os
Posted by Zheng Hacker 3 years, 5 months ago 
Zheng Yejian <zhengyejian1@huawei.com> 于2022年10月19日周三 09:56写道：
>
> On Wed, 19 Oct 2022 00:52:59 +0800, Zheng Wang <zyytlz.wz@163.com> wrote:
> > Gts may be freed in gru_check_chiplet_assignment.
> > The caller still use it after that, UAF happens.
> >
> > Fix it by introducing a return value to see if it's in error path or not.
> > Free the gts in caller if gru_check_chiplet_assignment check failed.
> >
> > Fixes: 55484c45dbec ("gru: allow users to specify gru chiplet 2")
> > Reported-by: Zheng Wang <hackerzheng666@gmail.com>
> > Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
> > ---
> > v3:
> > - add preempt_enable and use VM_FAULT_NOPAGE as failure code by Yejian
>
> Sorry, maybe I didn't make it clearly enough in
> https://lore.kernel.org/lkml/20221009201418.509417-1-zhengyejian1@huawei.com/
>
> What I suggested was that return value of gru_fault() should be some VM_FAULT_XXX,
> because this return value would be specially handled in pagefault flow like:
>     __do_fault() {
>         ......
>         ret = vma->vm_ops->fault(vmf);
>         ......
>     }
>
> But in other functions that are not in pagefault flow, common error code
> like -EINVAL is ok. Meanings of every error code can be found in:
> https://man7.org/linux/man-pages/man3/errno.3.html
>
> By the way, these are all personal suggestions. You need Reviewed-by Maintainer :)

Hi Zheng Yejian,

Thanks for your suggestion! Sorry for my late patch, I was a little
bit busy last week. I think the trigger path of this issue is divise.
So yes, -EINVAL may be a better choice.

I'll update the patch soon after.

Best Regards,
Zheng Wang
Re: [PATCH v3] misc: sgi-gru: fix use-after-free error in gru_set_context_option, gru_fault and gru_handle_user_call_os
Posted by Zheng Yejian 3 years, 5 months ago 
On Wed, 19 Oct 2022 10:39:11 +0800, Zheng Hacker <hackerzheng666@gmail.com> wrote:
> Zheng Yejian <zhengyejian1@huawei.com> 于2022年10月19日周三 09:56写道：
> > On Wed, 19 Oct 2022 00:52:59 +0800, Zheng Wang <zyytlz.wz@163.com> wrote:
> > > Gts may be freed in gru_check_chiplet_assignment.
> > > The caller still use it after that, UAF happens.
> > >
> > > Fix it by introducing a return value to see if it's in error path or not.
> > > Free the gts in caller if gru_check_chiplet_assignment check failed.
> > >
> > > Fixes: 55484c45dbec ("gru: allow users to specify gru chiplet 2")
> > > Reported-by: Zheng Wang <hackerzheng666@gmail.com>
> > > Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
> > > ---
> > > v3:
> > > - add preempt_enable and use VM_FAULT_NOPAGE as failure code by Yejian
> >
> > Sorry, maybe I didn't make it clearly enough in
> > https://lore.kernel.org/lkml/20221009201418.509417-1-zhengyejian1@huawei.com/
> >
> > What I suggested was that return value of gru_fault() should be some VM_FAULT_XXX,
> > because this return value would be specially handled in pagefault flow like:
> >     __do_fault() {
> >         ......
> >         ret = vma->vm_ops->fault(vmf);
> >         ......
> >     }
> >
> > But in other functions that are not in pagefault flow, common error code
> > like -EINVAL is ok. Meanings of every error code can be found in:
> > https://man7.org/linux/man-pages/man3/errno.3.html
> >
> > By the way, these are all personal suggestions. You need Reviewed-by Maintainer :)
>
> Hi Zheng Yejian,
>
> Thanks for your suggestion! Sorry for my late patch, I was a little
> bit busy last week. I think the trigger path of this issue is divise.
> So yes, -EINVAL may be a better choice.
>
> I'll update the patch soon after.

Hi Wang,
I'm very glad for my suggestion being adopted.
And really thank you for your efforts!

>
> Best Regards,
> Zheng Wang

Patchew 2.1-devel-ea86937

© 2016 - 2026 Red Hat, Inc.