linux-kernel - Re: [PATCH] drm/xe: Fix uninitialized pointer read in xe_vm_bind_kernel

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: <guahdobdjkqgkxiqmcabakdgtjsqhry7lfbfvd4yghbgqg66ps@u53z7leyikry>
Date: Thu, 6 Feb 2025 10:58:38 -0600
From: Lucas De Marchi <lucas.demarchi@...el.com>
To: Dheeraj Reddy Jonnalagadda <dheeraj.linuxdev@...il.com>
CC: <thomas.hellstrom@...ux.intel.com>, <rodrigo.vivi@...el.com>,
	<airlied@...il.com>, <simona@...ll.ch>, <intel-xe@...ts.freedesktop.org>,
	<dri-devel@...ts.freedesktop.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] drm/xe: Fix uninitialized pointer read in
 xe_vm_bind_kernel_bo

On Thu, Feb 06, 2025 at 07:17:06PM +0530, Dheeraj Reddy Jonnalagadda wrote:
>xe_vm_bind_kernel_bo() declares the variable fence without initializing
>it. If an error occurs before fence is assigned a valid value, the
>function may return an uninitialized pointer.
>
>Specifically, this can happen if:
>
>- vm_bind_ioctl_ops_create() fails, leading to release_vm_lock error
>  path

are you actually seeing an error or does this come from (static) analysis?

we are only on the error handling path when err is set to something.
Using your example:

	ops = vm_bind_ioctl_ops_create(vm, bo, 0, addr, bo->size,    
				       DRM_XE_VM_BIND_OP_MAP, 0, 0,  
				       vm->xe->pat.idx[cache_lvl]);  
	if (IS_ERR(ops)) {                                           
		err = PTR_ERR(ops);                                  
		goto release_vm_lock;
	}

all the other `goto ...`  before `fence = ...` assign err, too.
On the the error handling we do:

	if (err)
		fence = ERR_PTR(err);

So I'm not seeing the bug mentioned.

Lucas De Marchi

>
>Initialize fence to NULL at declaration.
>
>Fixes: dcdd6b84d9ac ("drm/xe/pxp: Allocate PXP execution resources")
>Signed-off-by: Dheeraj Reddy Jonnalagadda <dheeraj.linuxdev@...il.com>
>---
> drivers/gpu/drm/xe/xe_vm.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
>diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
>index d664f2e418b2..b6ca53ede904 100644
>--- a/drivers/gpu/drm/xe/xe_vm.c
>+++ b/drivers/gpu/drm/xe/xe_vm.c
>@@ -3184,7 +3184,7 @@ struct dma_fence *xe_vm_bind_kernel_bo(struct xe_vm *vm, struct xe_bo *bo,
> {
> 	struct xe_vma_ops vops;
> 	struct drm_gpuva_ops *ops = NULL;
>-	struct dma_fence *fence;
>+	struct dma_fence *fence = NULL;
> 	int err;
>
> 	xe_bo_get(bo);
>-- 
>2.34.1
>