Script 'mail_helper' called by obssrc
Hello community,
here is the log from the commit of package nvidia-open-driver-G06-signed for
openSUSE:Factory checked in at 2025-07-02 12:10:46
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Comparing /work/SRC/openSUSE:Factory/nvidia-open-driver-G06-signed (Old)
and /work/SRC/openSUSE:Factory/.nvidia-open-driver-G06-signed.new.7067
(New)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "nvidia-open-driver-G06-signed"
Wed Jul 2 12:10:46 2025 rev:71 rq:1289518 version:570.169
Changes:
--------
---
/work/SRC/openSUSE:Factory/nvidia-open-driver-G06-signed/nvidia-open-driver-G06-signed.changes
2025-06-18 16:03:04.441359749 +0200
+++
/work/SRC/openSUSE:Factory/.nvidia-open-driver-G06-signed.new.7067/nvidia-open-driver-G06-signed.changes
2025-07-02 12:13:55.757773650 +0200
@@ -1,0 +2,9 @@
+Tue Jun 24 09:47:53 UTC 2025 - Stefan Dirsch <[email protected]>
+
+- 0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch
+ 0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch
+ * buildfixes against Kernel 6.16 picked up from
+ https://github.com/CachyOS/CachyOS-PKGBUILDS.git
+ --> nvidia/nvidia-utils
+
+-------------------------------------------------------------------
New:
----
0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch
0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch
----------(New B)----------
New:
- 0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch
0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch
New:- 0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch
0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch
* buildfixes against Kernel 6.16 picked up from
----------(New E)----------
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Other differences:
------------------
++++++ nvidia-open-driver-G06-signed.spec ++++++
--- /var/tmp/diff_new_pack.gTBnNV/_old 2025-07-02 12:13:56.521805191 +0200
+++ /var/tmp/diff_new_pack.gTBnNV/_new 2025-07-02 12:13:56.521805191 +0200
@@ -82,6 +82,8 @@
Source12: pesign-spec-macros
Source14: group-source-files.pl
Source15: kmp-trigger.sh
+Patch0: 0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch
+Patch1: 0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch
BuildRequires: %{kernel_module_package_buildreqs}
BuildRequires: fdupes
BuildRequires: gcc-c++
++++++ 0003-nv-dmabuf-Inline-dma_buf_attachment_is_dynamic.patch ++++++
>From eb769c9dac0488266e40cddb1e05e7049028125c Mon Sep 17 00:00:00 2001
From: Eric Naim <[email protected]>
Date: Mon, 9 Jun 2025 13:00:54 +0700
Subject: [PATCH 4/4] nv-dmabuf: Inline dma_buf_attachment_is_dynamic()
6.16 made dma_buf_attachment_is_dynamic() an internal function, meaning
that modules like nvidia cannot use it now. This is not much of a
problem since we can inline the function body directly.
Signed-off-by: Eric Naim <[email protected]>
---
kernel-open/nvidia/nv-dmabuf.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/kernel-open/nvidia/nv-dmabuf.c b/kernel-open/nvidia/nv-dmabuf.c
index 8c1447b90a01..1c77ebe243a1 100644
--- a/kernel-open/nvidia/nv-dmabuf.c
+++ b/kernel-open/nvidia/nv-dmabuf.c
@@ -837,12 +837,10 @@ nv_dma_buf_map(
// PCIe mapping, importers must be able to handle peer MMIO resources
// not backed by struct page.
//
-#if defined(NV_DMA_BUF_HAS_DYNAMIC_ATTACHMENT) && \
- defined(NV_DMA_BUF_ATTACHMENT_HAS_PEER2PEER)
+#if defined(NV_DMA_BUF_ATTACHMENT_HAS_PEER2PEER)
if (((!priv->nv->coherent) ||
(priv->mapping_type == NV_DMABUF_EXPORT_MAPPING_TYPE_FORCE_PCIE)) &&
- dma_buf_attachment_is_dynamic(attachment) &&
- !attachment->peer2peer)
+ !!attachment->importer_ops && !attachment->peer2peer)
{
nv_printf(NV_DBG_ERRORS,
"NVRM: failed to map dynamic attachment with no P2P
support\n");
--
2.49.0
++++++ 0004-nvidia-uvm-Disable-SVA-support-for-6.16.patch ++++++
>From 721c81877d45bca7627e07030b01b22365e7055a Mon Sep 17 00:00:00 2001
From: Eric Naim <[email protected]>
Date: Mon, 9 Jun 2025 13:26:58 +0700
Subject: [PATCH 5/5] nvidia-uvm: Disable SVA support for 6.16+
6.16 removed IOMMU_DEV_FEAT_SVA because it is unused. Until there is a
proper fix from NVIDIA, disable this feature for now.
Signed-off-by: Eric Naim <[email protected]>
---
kernel-open/nvidia-uvm/uvm_ats_sva.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/kernel-open/nvidia-uvm/uvm_ats_sva.h
b/kernel-open/nvidia-uvm/uvm_ats_sva.h
index 293c563b4bba..a30f4ff7186f 100644
--- a/kernel-open/nvidia-uvm/uvm_ats_sva.h
+++ b/kernel-open/nvidia-uvm/uvm_ats_sva.h
@@ -28,6 +28,7 @@
#include "uvm_forward_decl.h"
#include <linux/iommu.h>
+#include <linux/version.h>
// For ATS support on aarch64, arm_smmu_sva_bind() is needed for
// iommu_sva_bind_device() calls. Unfortunately, arm_smmu_sva_bind() is not
@@ -44,7 +45,7 @@
// 701fac40384f07197b106136012804c3cae0b3de (02/15/2022) removed ioasid_get()
// and added mm_pasid_drop().
#if UVM_CAN_USE_MMU_NOTIFIERS() && (defined(NV_IOASID_GET_PRESENT) ||
defined(NV_MM_PASID_DROP_PRESENT))
- #if defined(CONFIG_IOMMU_SVA)
+ #if defined(CONFIG_IOMMU_SVA) && LINUX_VERSION_CODE <
KERNEL_VERSION(6, 16, 0)
#define UVM_ATS_SVA_SUPPORTED() 1
#else
#define UVM_ATS_SVA_SUPPORTED() 0
--
2.49.0
