Greetings Greg and others,
I am sending a two part patch fix to prove to Greg I have changed.
Below this message are the patches for a bug fix in the gma500 driver
code. Unfortunately I haven't gotten a maintainer acknowledge and
therefore it's not in the merge for the 4.1 release.
Nick
Patch 1
1 From 2d2ddb5d9a2c4fcbae45339d4f775fcde49f36e1 Mon Sep 17 00:00:00 2001
2 From: Nicholas Krause <xerofoify@xxxxxxxxx>
3 Date: Wed, 13 May 2015 21:36:44 -0400
4 Subject: [PATCH 1/2] gma500:Add proper use of the variable ret for the
5 function, psb_mmu_inset_pfn_sequence
6
7 This adds proper use of the variable ret by returning it
8 at the end of the function, psb_mmu_inset_pfn_sequence for
9 indicating to callers when an error has occurred. Further
10 more remove the unneeded double setting of ret to the error
11 code, -ENOMEM after checking if a call to the function,
12 psb_mmu_pt_alloc_map_lock is successful.
13
14 Signed-off-by: Nicholas Krause <xerofoify@xxxxxxxxx>
15 ---
16 drivers/gpu/drm/gma500/mmu.c | 7 +++----
17 1 file changed, 3 insertions(+), 4 deletions(-)
18
19 diff --git a/drivers/gpu/drm/gma500/mmu.c b/drivers/gpu/drm/gma500/mmu.c
20 index 0eaf11c..d2c4bac 100644
21 --- a/drivers/gpu/drm/gma500/mmu.c
22 +++ b/drivers/gpu/drm/gma500/mmu.c
23 @@ -677,10 +677,9 @@ int psb_mmu_insert_pfn_sequence(struct psb_mmu_pd *pd, uint32_t start_pfn,
24 do {
25 next = psb_pd_addr_end(addr, end);
26 pt = psb_mmu_pt_alloc_map_lock(pd, addr);
27 - if (!pt) {
28 - ret = -ENOMEM;
29 + if (!pt)
30 goto out;
31 - }
32 +
33 do {
34 pte = psb_mmu_mask_pte(start_pfn++, type);
35 psb_mmu_set_pte(pt, addr, pte);
36 @@ -700,7 +699,7 @@ out:
37 if (pd->hw_context != -1)
38 psb_mmu_flush(pd->driver);
39
40 - return 0;
41 + return ret;
42 }
43
44 int psb_mmu_insert_pages(struct psb_mmu_pd *pd, struct page **pages,
45 --
46 2.1.4
47
Patch 2
1 From f1802ff61ef69ff9ecaaeadb941d4ed725a0255a Mon Sep 17 00:00:00 2001
2 From: Nicholas Krause <xerofoify@xxxxxxxxx>
3 Date: Fri, 22 May 2015 23:10:49 -0400
4 Subject: [PATCH 2/2] gma500: Add new error checking for the function, psb_driver_load
5
6 This adds new error checking to the function, psb_driver_load
7 for when the function, psb_mmu_inset_pfn_sequence fails to grab
8 memory for the internal function call to psb_pd_addr_end. In
9 addition we now implement this by checking for a error code
10 return value from the internal call to the function,
11 psb_mmu_inset_pfn_sequence for the function, psb_driver_load.
12 If a error code is returned we must jump to a new label,
13 unlock_err in order to deallocate our usage count by
14 one for the usage of the semaphore, sem before unloading
15 the drm_device structure and returning a error code.
16
17 Signed-off-by: Nicholas Krause <xerofoify@xxxxxxxxx>
18 ---
19 drivers/gpu/drm/gma500/psb_drv.c | 4 ++++
20 1 file changed, 4 insertions(+)
21
22 diff --git a/drivers/gpu/drm/gma500/psb_drv.c b/drivers/gpu/drm/gma500/psb_drv.c
23 index 92e7e57..4cef2c9 100644
24 --- a/drivers/gpu/drm/gma500/psb_drv.c
25 +++ b/drivers/gpu/drm/gma500/psb_drv.c
26 @@ -343,6 +343,8 @@ static int psb_driver_load(struct drm_device *dev, unsigned long flags)
27 dev_priv->stolen_base >> PAGE_SHIFT,
28 pg->gatt_start,
29 pg->stolen_size >> PAGE_SHIFT, 0);
30 + if (ret)
31 + goto unlock_err;
32 up_read(&pg->sem);
33
34 psb_mmu_set_pd_context(psb_mmu_get_default_pd(dev_priv->mmu), 0);
35 @@ -405,6 +407,8 @@ static int psb_driver_load(struct drm_device *dev, unsigned long flags)
36 #endif
37 /* Intel drm driver load is done, continue doing pvr load */
38 return 0;
39 +unlock_err:
40 + up_read(&pg->sem);
41 out_err:
42 psb_driver_unload(dev);
43 return ret;
44 --
45 2.1.4
46
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
