Received-SPF: pass (google.com: domain of linux-kernel+bounces-93763-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
From: Li Zhijian <lizhijian@fujitsu.com>
To: linux-kernel@vger.kernel.org
Cc: y-goto@fujitsu.com,
	Alison Schofield <alison.schofield@intel.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Baoquan He <bhe@redhat.com>,
	Borislav Petkov <bp@alien8.de>,
	Dan Williams <dan.j.williams@intel.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	Dave Jiang <dave.jiang@intel.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	hpa@zytor.com,
	Ingo Molnar <mingo@redhat.com>,
	Ira Weiny <ira.weiny@intel.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Vishal Verma <vishal.l.verma@intel.com>,
	linux-cxl@vger.kernel.org,
	linux-mm@kvack.org,
	nvdimm@lists.linux.dev,
	x86@kernel.org,
	kexec@lists.infradead.org,
	Li Zhijian <lizhijian@fujitsu.com>
Subject: [PATCH v3 7/7] nvdimm: set force_raw=1 in kdump kernel
Date: Wed,  6 Mar 2024 18:28:46 +0800
Message-Id: <20240306102846.1020868-8-lizhijian@fujitsu.com>
In-Reply-To: <20240306102846.1020868-1-lizhijian@fujitsu.com>
References: <20240306102846.1020868-1-lizhijian@fujitsu.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

The virtually mapped memory map allows storing struct page objects for
persistent memory devices in pre-allocated storage on those devices.
These 'struct page objects' on devices are also known as metadata.

During libnvdimm/nd_pmem are loading, the previous metadata will
be re-constructed to fit the current running kernel. For kdump purpose,
these metadata should not be touched until the dumping is done so that
the metadata is identical.

To achieve this, we have some options
1. Don't provide libnvdimm driver in kdump kernel rootfs/initramfs
2. Disable libnvdimm driver by specific comline parameters:
   (initcall_blacklist=libnvdimm_init libnvdimm.blacklist=1 rd.driver.blacklist=libnvdimm)
3. Enforce force_raw=1 for nvdimm namespace, because when force_raw=1,
   metadata will not be re-constructed again. This may also result in
   the pmem doesn't work before a few extra configurations.

Here apply the 3rd option.

CC: Dan Williams <dan.j.williams@intel.com>
CC: Vishal Verma <vishal.l.verma@intel.com>
CC: Dave Jiang <dave.jiang@intel.com>
CC: Ira Weiny <ira.weiny@intel.com>
CC: nvdimm@lists.linux.dev
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
---
 drivers/nvdimm/namespace_devs.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c
index d6d558f94d6b..04f855c7f0b1 100644
--- a/drivers/nvdimm/namespace_devs.c
+++ b/drivers/nvdimm/namespace_devs.c
@@ -9,6 +9,7 @@
 #include <linux/slab.h>
 #include <linux/list.h>
 #include <linux/nd.h>
+#include <linux/crash_dump.h>
 #include "nd-core.h"
 #include "pmem.h"
 #include "pfn.h"
@@ -1513,6 +1514,8 @@ struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev)
 			return ERR_PTR(-ENODEV);
 	}
 
+	if (is_kdump_kernel())
+		ndns->force_raw = true;
 	return ndns;
 }
 EXPORT_SYMBOL(nvdimm_namespace_common_probe);
-- 
2.29.2