Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Zheng Wang <zyytlz.wz@163.com>
To:     zhi.a.wang@intel.com
Cc:     1002992920@qq.com, airlied@gmail.com, airlied@linux.ie,
        alex000young@gmail.com, dri-devel@lists.freedesktop.org,
        gregkh@linuxfoundation.org, hackerzheng666@gmail.com,
        intel-gfx@lists.freedesktop.org,
        intel-gvt-dev@lists.freedesktop.org,
        joonas.lahtinen@linux.intel.com, linux-kernel@vger.kernel.org,
        security@kernel.org, tvrtko.ursulin@linux.intel.com,
        zhenyuw@linux.intel.com, zyytlz.wz@163.com
Subject: [PATCH v6] drm/i915/gvt: fix double free bug in split_2MB_gtt_entry
Date:   Fri, 30 Dec 2022 00:56:41 +0800
Message-Id: <20221229165641.1192455-1-zyytlz.wz@163.com>
In-Reply-To: <11728bc1-7b59-1623-b517-d1a0d57eb275@intel.com>
References: <11728bc1-7b59-1623-b517-d1a0d57eb275@intel.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

If intel_gvt_dma_map_guest_page failed, it will call
 ppgtt_invalidate_spt, which will finally free the spt.
 But the caller function ppgtt_populate_spt_by_guest_entry
 does not notice that, it will free spt again in its error
 path.

Fix this by canceling the mapping of DMA address and freeing sub_spt.
Besides, leave the handle of spt destroy to caller function instead
of callee function when error occurs.

Fixes: b901b252b6cf ("drm/i915/gvt: Add 2M huge gtt support")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Reviewed-by: Zhenyu Wang <zhenyuw@linux.intel.com>
---
v6:
- remove the code for setting unused variable to NULL and fix type suggested
by Zhenyu

v5:
- remove unnecessary switch-case code for there is only one particular case,
correct the unmap target from parent_spt to sub_spt.add more details in
commit message. All suggested by Zhenyu

v4:
- fix by undo the mapping of DMA address and free sub_spt suggested by Zhi

v3:
- correct spelling mistake and remove unused variable suggested by Greg

v2: https://lore.kernel.org/all/20221006165845.1735393-1-zyytlz.wz@163.com/

v1: https://lore.kernel.org/all/20220928033340.1063949-1-zyytlz.wz@163.com/
---
 drivers/gpu/drm/i915/gvt/gtt.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
index 51e5e8fb505b..7379e8d98417 100644
--- a/drivers/gpu/drm/i915/gvt/gtt.c
+++ b/drivers/gpu/drm/i915/gvt/gtt.c
@@ -1209,10 +1209,8 @@ static int split_2MB_gtt_entry(struct intel_vgpu *vgpu,
 	for_each_shadow_entry(sub_spt, &sub_se, sub_index) {
 		ret = intel_gvt_dma_map_guest_page(vgpu, start_gfn + sub_index,
 						   PAGE_SIZE, &dma_addr);
-		if (ret) {
-			ppgtt_invalidate_spt(spt);
-			return ret;
-		}
+		if (ret)
+			goto err;
 		sub_se.val64 = se->val64;
 
 		/* Copy the PAT field from PDE. */
@@ -1231,6 +1229,17 @@ static int split_2MB_gtt_entry(struct intel_vgpu *vgpu,
 	ops->set_pfn(se, sub_spt->shadow_page.mfn);
 	ppgtt_set_shadow_entry(spt, se, index);
 	return 0;
+err:
+	/* Cancel the existing addess mappings of DMA addr. */
+	for_each_present_shadow_entry(sub_spt, &sub_se, sub_index) {
+		gvt_vdbg_mm("invalidate 4K entry\n");
+		ppgtt_invalidate_pte(sub_spt, &sub_se);
+	}
+	/* Release the new allocated spt. */
+	trace_spt_change(sub_spt->vgpu->id, "release", sub_spt,
+		sub_spt->guest_page.gfn, sub_spt->shadow_page.type);
+	ppgtt_free_spt(sub_spt);
+	return ret;
 }
 
 static int split_64KB_gtt_entry(struct intel_vgpu *vgpu,
-- 
2.25.1