Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Masayoshi Mizuma <msys.mizuma@gmail.com>
To:     Borislav Petkov <bp@alien8.de>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>,
        "H. Peter Anvin" <hpa@zytor.com>, x86@kernel.org,
        Baoquan He <bhe@redhat.com>
Cc:     Masayoshi Mizuma <msys.mizuma@gmail.com>,
        Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>,
        linux-kernel@vger.kernel.org
Subject: [PATCH v4 4/4] x86/mm/KASLR: Adjust the padding size for the direct  mapping.
Date:   Fri,  1 Nov 2019 21:09:11 -0400
Message-Id: <20191102010911.21460-5-msys.mizuma@gmail.com>
In-Reply-To: <20191102010911.21460-1-msys.mizuma@gmail.com>
References: <20191102010911.21460-1-msys.mizuma@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

From: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>

The system sometimes crashes while memory hot-adding on KASLR
enabled system. The crash happens because the regions pointed by
kaslr_regions[].base are overwritten by the hot-added memory.

It happens because of the padding size for kaslr_regions[].base isn't
enough for the system whose physical memory layout has huge space for
memory hotplug. kaslr_regions[].base points "actual installed
memory size + padding" or higher address. So, if the "actual + padding"
is lower address than the maximum memory address, which means the memory
address reachable by memory hot-add, kaslr_regions[].base is destroyed by
the overwritten.

  address
    ^
    |------- maximum memory address (Hotplug)
    |                                    ^
    |------- kaslr_regions[0].base       | Hotadd-able region
    |     ^                              |
    |     | padding                      |
    |     V                              V
    |------- actual memory address (Installed on boot)
    |

Fix it by getting the maximum memory address from SRAT and store
the value in boot_param, then set the padding size while KASLR
initializing if the default padding size isn't enough.

Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
---
 arch/x86/mm/kaslr.c | 65 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 50 insertions(+), 15 deletions(-)

diff --git a/arch/x86/mm/kaslr.c b/arch/x86/mm/kaslr.c
index dc6182eec..a80eed563 100644
--- a/arch/x86/mm/kaslr.c
+++ b/arch/x86/mm/kaslr.c
@@ -70,15 +70,60 @@ static inline bool kaslr_memory_enabled(void)
 	return kaslr_enabled() && !IS_ENABLED(CONFIG_KASAN);
 }
 
+/*
+ * Even though a huge virtual address space is reserved for the direct
+ * mapping of physical memory, e.g in 4-level paging mode, it's 64TB,
+ * rare system can own enough physical memory to use it up, most are
+ * even less than 1TB. So with KASLR enabled, we adapt the size of
+ * direct mapping area to the size of actual physical memory plus the
+ * configured padding CONFIG_RANDOMIZE_MEMORY_PHYSICAL_PADDING.
+ * The left part will be taken out to join memory randomization.
+ */
+static inline unsigned long calc_direct_mapping_size(void)
+{
+	unsigned long padding = CONFIG_RANDOMIZE_MEMORY_PHYSICAL_PADDING;
+	unsigned long size_tb, memory_tb;
+#ifdef CONFIG_MEMORY_HOTPLUG
+	unsigned long actual, maximum, base;
+
+	if (boot_params.max_addr) {
+		/*
+		 * The padding size should set to get for kaslr_regions[].base
+		 * bigger address than the maximum memory address the system can
+		 * have. kaslr_regions[].base points "actual size + padding" or
+		 * higher address. If "actual size + padding" points the lower
+		 * address than the maximum memory size, fix the padding size.
+		 */
+		actual = roundup(PFN_PHYS(max_pfn), 1UL << TB_SHIFT);
+		maximum = roundup(boot_params.max_addr, 1UL << TB_SHIFT);
+		base = actual + (padding << TB_SHIFT);
+
+		if (maximum > base)
+			padding = (maximum - actual) >> TB_SHIFT;
+	}
+#endif
+	memory_tb =  DIV_ROUND_UP(max_pfn << PAGE_SHIFT, 1UL << TB_SHIFT) +
+			padding;
+
+	size_tb = 1 << (MAX_PHYSMEM_BITS - TB_SHIFT);
+
+	/*
+	 * Adapt physical memory region size based on available memory
+	 */
+	if (memory_tb < size_tb)
+		size_tb = memory_tb;
+
+	return size_tb;
+}
+
 /* Initialize base and padding for each memory region randomized with KASLR */
 void __init kernel_randomize_memory(void)
 {
-	size_t i;
-	unsigned long vaddr_start, vaddr;
-	unsigned long rand, memory_tb;
-	struct rnd_state rand_state;
+	unsigned long vaddr_start, vaddr, rand;
 	unsigned long remain_entropy;
 	unsigned long vmemmap_size;
+	struct rnd_state rand_state;
+	size_t i;
 
 	vaddr_start = pgtable_l5_enabled() ? __PAGE_OFFSET_BASE_L5 : __PAGE_OFFSET_BASE_L4;
 	vaddr = vaddr_start;
@@ -95,20 +140,10 @@ void __init kernel_randomize_memory(void)
 	if (!kaslr_memory_enabled())
 		return;
 
-	kaslr_regions[0].size_tb = 1 << (MAX_PHYSMEM_BITS - TB_SHIFT);
+	kaslr_regions[0].size_tb = calc_direct_mapping_size();
 	kaslr_regions[1].size_tb = VMALLOC_SIZE_TB;
 
-	/*
-	 * Update Physical memory mapping to available and
-	 * add padding if needed (especially for memory hotplug support).
-	 */
 	BUG_ON(kaslr_regions[0].base != &page_offset_base);
-	memory_tb = DIV_ROUND_UP(max_pfn << PAGE_SHIFT, 1UL << TB_SHIFT) +
-		CONFIG_RANDOMIZE_MEMORY_PHYSICAL_PADDING;
-
-	/* Adapt phyiscal memory region size based on available memory */
-	if (memory_tb < kaslr_regions[0].size_tb)
-		kaslr_regions[0].size_tb = memory_tb;
 
 	/*
 	 * Calculate the vmemmap region size in TBs, aligned to a TB
-- 
2.20.1