[gentoo-user] [PATCH] Linus breaks nvidia-drivers again

[gentoo-user] [PATCH] Linus breaks nvidia-drivers again

[walt]

[-- Attachment #1: Type: text/plain, Size: 179 bytes --]

Well, this is really a temporary workaround for people who like to
run the latest git kernel from Linus, and it's only tested for
nvidia-drivers-302.17-r1, and only on ~amd_64.



[-- Attachment #2: nvidia.patch --]
[-- Type: text/x-patch, Size: 1995 bytes --]

diff -ur NVIDIA-Linux-x86_64-302.17.orig/kernel/conftest.sh NVIDIA-Linux-x86_64-302.17/kernel/conftest.sh
--- NVIDIA-Linux-x86_64-302.17.orig/kernel/conftest.sh	2012-06-12 16:03:27.000000000 -0700
+++ NVIDIA-Linux-x86_64-302.17/kernel/conftest.sh	2012-10-15 08:12:56.594959000 -0700
@@ -126,7 +126,7 @@
     CFLAGS="$BASE_CFLAGS $MACH_CFLAGS $OUTPUT_CFLAGS -I$HEADERS $AUTOCONF_CFLAGS"
 
     if [ "$ARCH" = "i386" -o "$ARCH" = "x86_64" ]; then
-        CFLAGS="$CFLAGS -I$SOURCES/arch/x86/include -I$OUTPUT/arch/x86/include/generated"
+        CFLAGS="$CFLAGS -I$SOURCES/arch/x86/include -I$OUTPUT/arch/x86/include/generated -I$SOURCES/include/uapi"
     elif [ "$ARCH" = "arm" ]; then
         CFLAGS="$CFLAGS -I$SOURCES/arch/arm/include -I$OUTPUT/arch/arm/include/generated"
     fi
diff -ur NVIDIA-Linux-x86_64-302.17.orig/kernel/nv-acpi.c NVIDIA-Linux-x86_64-302.17/kernel/nv-acpi.c
--- NVIDIA-Linux-x86_64-302.17.orig/kernel/nv-acpi.c	2012-06-12 16:03:27.000000000 -0700
+++ NVIDIA-Linux-x86_64-302.17/kernel/nv-acpi.c	2012-10-15 08:15:15.888959000 -0700
@@ -300,7 +300,7 @@
     if (pNvAcpiObject->notify_handler_installed)
     {
         // no status returned for this function
-        acpi_os_wait_events_complete(NULL);
+        acpi_os_wait_events_complete();
 
         // remove event notifier
         status = acpi_remove_notify_handler(device->handle, ACPI_DEVICE_NOTIFY, nv_acpi_event);
diff -ur NVIDIA-Linux-x86_64-302.17.orig/kernel/nv-mmap.c NVIDIA-Linux-x86_64-302.17/kernel/nv-mmap.c
--- NVIDIA-Linux-x86_64-302.17.orig/kernel/nv-mmap.c	2012-06-12 16:03:27.000000000 -0700
+++ NVIDIA-Linux-x86_64-302.17/kernel/nv-mmap.c	2012-10-15 08:17:13.084959000 -0700
@@ -450,7 +450,7 @@
         NV_PRINT_AT(NV_DBG_MEMINFO, at);
         nv_vm_list_page_count(&at->page_table[i], pages);
 
-        vma->vm_flags |= (VM_IO | VM_LOCKED | VM_RESERVED);
+        vma->vm_flags |= (VM_IO | VM_LOCKED | VM_IO);
 
 #if defined(VM_DRIVER_PAGES)
         vma->vm_flags |= VM_DRIVER_PAGES;

[gentoo-user] Re: [PATCH] Linus breaks ati-drivers again

[walt]

[-- Attachment #1: Type: text/plain, Size: 179 bytes --]


Well, this is really a temporary workaround for people who like to
run the latest git kernel from Linus, and it's only tested for
ati-drivers-12.9_beta, and only on ~amd_64.
  


[-- Attachment #2: ati.patch --]
[-- Type: text/x-patch, Size: 3601 bytes --]

--- common/lib/modules/fglrx/build_mod/firegl_public.orig	2012-10-15 10:10:58.593454377 -0700
+++ common/lib/modules/fglrx/build_mod/firegl_public.c	2012-10-15 10:12:56.453972670 -0700
@@ -3892,7 +3892,7 @@
                 KCL_DEBUG_ERROR(REMAP_PAGE_RANGE_STR " failed\n");
                 return -EAGAIN;
             }
-            vma->vm_flags |= VM_SHM | VM_RESERVED; /* Don't swap */
+            vma->vm_flags |= VM_SHM | VM_IO; /* Don't swap */
             vma->vm_ops = &vm_ops;
 			break;
 
@@ -3922,14 +3922,14 @@
                 KCL_DEBUG_ERROR(REMAP_PAGE_RANGE_STR " failed\n");
                 return -EAGAIN;
             }
-            vma->vm_flags |= VM_SHM | VM_RESERVED; /* Don't swap */
+            vma->vm_flags |= VM_SHM | VM_IO; /* Don't swap */
             vma->vm_ops = &vm_ops;
             }
 			break;
 #endif                    
 
         case __KE_SHM:
-            vma->vm_flags |= VM_SHM | VM_RESERVED; /* Don't swap */
+            vma->vm_flags |= VM_SHM | VM_IO; /* Don't swap */
             vma->vm_ops = &vm_shm_ops;
             break;
 
@@ -3937,7 +3937,7 @@
 
             pages = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
 
-            vma->vm_flags |= VM_RESERVED;
+            vma->vm_flags |= VM_IO;
 
             //vma->vm_flags |=  VM_SHM | VM_LOCKED; /* DDDDDDDDDDon't swap */
             //vma->vm_mm->locked_vm += pages; /* Kernel tracks aqmount of locked pages */
@@ -3946,14 +3946,14 @@
 
         case __KE_CTX:
             pages = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-            vma->vm_flags |= VM_LOCKED | VM_SHM | VM_RESERVED; /* Don't swap */
+            vma->vm_flags |= VM_LOCKED | VM_SHM | VM_IO; /* Don't swap */
             vma->vm_mm->locked_vm += pages; /* Kernel tracks aqmount of locked pages */
             vma->vm_ops = &vm_ctx_ops;
             break;
 
         case __KE_PCI_BQS:
             pages = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-            vma->vm_flags |= VM_LOCKED | VM_SHM | VM_RESERVED; /* Don't swap */
+            vma->vm_flags |= VM_LOCKED | VM_SHM | VM_IO; /* Don't swap */
             vma->vm_mm->locked_vm += pages; /* Kernel tracks aqmount of locked pages */
             vma->vm_ops = &vm_pci_bq_ops;
             break;
@@ -3984,9 +3984,9 @@
                     return -EAGAIN;
                 }
 #ifdef __x86_64__
-                vma->vm_flags |= VM_RESERVED;
+                vma->vm_flags |= VM_IO;
 #else
-                vma->vm_flags |= VM_SHM | VM_RESERVED; /* Don't swap */
+                vma->vm_flags |= VM_SHM | VM_IO; /* Don't swap */
 #endif
                 vma->vm_ops = &vm_ops;
             }
@@ -4015,9 +4015,9 @@
                     return -EAGAIN;
                 }
 #ifdef __x86_64__
-                vma->vm_flags |= VM_RESERVED;
+                vma->vm_flags |= VM_IO;
 #else
-                vma->vm_flags |= VM_SHM | VM_RESERVED; /* Don't swap */
+                vma->vm_flags |= VM_SHM | VM_IO; /* Don't swap */
 #endif
                 vma->vm_ops = &vm_agp_bq_ops;
             }
@@ -4025,7 +4025,7 @@
 #endif /* __AGP__BUILTIN__ */
 
         case __KE_KMAP:
-		    vma->vm_flags |= VM_SHM | VM_RESERVED;
+		    vma->vm_flags |= VM_SHM | VM_IO;
             vma->vm_ops = &vm_kmap_ops;
             if (readonly && (vma->vm_flags & VM_WRITE))
             {
@@ -4046,7 +4046,7 @@
 #endif            
             // fall through
          case __KE_GART_CACHEABLE:
-             vma->vm_flags |= VM_RESERVED;
+             vma->vm_flags |= VM_IO;
              vma->vm_ops = &vm_gart_ops;
              break;
         default:

[gentoo-user] Re: [PATCH] Linus breaks virtualbox-modules again

[walt]

[-- Attachment #1: Type: text/plain, Size: 183 bytes --]


Well, this is really a temporary workaround for people who like to
run the latest git kernel from Linus, and it's only tested for
virtualbox-modules-4.2.0-r1, and only on ~amd_64.



[-- Attachment #2: vbox.patch --]
[-- Type: text/x-patch, Size: 836 bytes --]

--- vboxdrv/r0drv/linux/memobj-r0drv-linux.c.orig	2012-06-12 01:08:34.000000000 -0700
+++ vboxdrv/r0drv/linux/memobj-r0drv-linux.c	2012-10-15 10:23:45.471813411 -0700
@@ -1457,7 +1457,7 @@
 
 #if   defined(VBOX_USE_INSERT_PAGE) && LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 22)
                     rc = vm_insert_page(vma, ulAddrCur, pMemLnxToMap->apPages[iPage]);
-                    vma->vm_flags |= VM_RESERVED; /* This flag helps making 100% sure some bad stuff wont happen (swap, core, ++). */
+                    vma->vm_flags |= VM_IO; /* This flag helps making 100% sure some bad stuff wont happen (swap, core, ++). */
 #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 11)
                     rc = remap_pfn_range(vma, ulAddrCur, page_to_pfn(pMemLnxToMap->apPages[iPage]), PAGE_SIZE, fPg);
 #elif defined(VBOX_USE_PAE_HACK)

Re: [gentoo-user] [PATCH] Linus breaks nvidia-drivers again

[Walter Dnes]

On Mon, Oct 15, 2012 at 09:05:27AM -0700, walt wrote
> Well, this is really a temporary workaround for people who like to
> run the latest git kernel from Linus, and it's only tested for
> nvidia-drivers-302.17-r1, and only on ~amd_64.

  Which kernel version does the breakage start at, so I know not to rush
into it when it goes mainstream?  Also, is there a walkthrough for
switching from Nvidia to Nouveau drivers?  I couldn't get Nouveau
working when I first tried, which is why I'm using the Nvidia driver
right now.  lspci shows...

01:00.0 VGA compatible controller: NVIDIA Corporation GT218 [GeForce 210] (rev a2) (prog-if 00 [VGA controller])
        Subsystem: ASUSTeK Computer Inc. Device 83c7

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-user] [PATCH] Linus breaks nvidia-drivers again

[pk]

On 2012-10-15 20:10, Walter Dnes wrote:
>   Which kernel version does the breakage start at, so I know not to rush
> into it when it goes mainstream?  Also, is there a walkthrough for
> switching from Nvidia to Nouveau drivers?  I couldn't get Nouveau
> working when I first tried, which is why I'm using the Nvidia driver
> right now.  lspci shows...
> 
> 01:00.0 VGA compatible controller: NVIDIA Corporation GT218 [GeForce 210] (rev a2) (prog-if 00 [VGA controller])
>         Subsystem: ASUSTeK Computer Inc. Device 83c7

Stupid question perhaps but...
...have you tried http://en.gentoo-wiki.com/wiki/Nouveau ?

There's also a link to a nvidia-to-noveau at the bottom of this link...

Best regards

Peter K

Re: [gentoo-user] [PATCH] Linus breaks nvidia-drivers again

[Walter Dnes]

On Mon, Oct 15, 2012 at 08:19:51PM +0200, pk wrote
> 
> Stupid question perhaps but...
> ...have you tried http://en.gentoo-wiki.com/wiki/Nouveau ?
> 
> There's also a link to a nvidia-to-noveau at the bottom of this link...

  Thanks.  I'll try the nvidia-to-noveau migration as shown in the
bottom link.

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

[gentoo-user] Re: [PATCH] Linus breaks nvidia-drivers again

[walt]

On 10/15/2012 11:10 AM, Walter Dnes wrote:
> On Mon, Oct 15, 2012 at 09:05:27AM -0700, walt wrote
>> >Well, this is really a temporary workaround for people who like to
>> >run the latest git kernel from Linus, and it's only tested for
>> >nvidia-drivers-302.17-r1, and only on ~amd_64.

>    Which kernel version does the breakage start at, so I know not to rush
> into it when it goes mainstream?

Linus committed the breakage to his git repository just last week, so the
nvidia people should have it fixed for 3.7.  Same goes for virtualbox and
ati-drivers, I hope.

But to answer your question, the breakage starts with linux-3.7.0-rc1.

Re: [gentoo-user] Re: [PATCH] Linus breaks nvidia-drivers again

[microcai]

change VM_RESERVERD to VM_DONTDUMP|VM_IO

2012/10/16 walt <w41ter@gmail.com>:
> On 10/15/2012 11:10 AM, Walter Dnes wrote:
>>
>> On Mon, Oct 15, 2012 at 09:05:27AM -0700, walt wrote
>>>
>>> >Well, this is really a temporary workaround for people who like to
>>> >run the latest git kernel from Linus, and it's only tested for
>>> >nvidia-drivers-302.17-r1, and only on ~amd_64.
>
>
>>    Which kernel version does the breakage start at, so I know not to rush
>> into it when it goes mainstream?
>
>
> Linus committed the breakage to his git repository just last week, so the
> nvidia people should have it fixed for 3.7.  Same goes for virtualbox and
> ati-drivers, I hope.
>
> But to answer your question, the breakage starts with linux-3.7.0-rc1.
>