arm64: add dump_user_range() to machine check safe

hulk inclusion
category: feature
bugzilla: https://gitee.com/openeuler/kernel/issues/I5GB28
CVE: NA

-------------------------------

In the dump_user_range() processing, the data of the user process is
dump to corefile, when hardware memory error is encountered during dump,
only the relevant processes are affected, so killing the user process and
isolate the user page with hardware memory errors is a more reasonable
choice than kernel panic.

The dump_user_range() typical usage scenarios is coredump. Coredump file
writing to fs is related to the specific implementation of fs's write_iter
operation. This patch only supports two typical fs write function
(_copy_from_iter/iov_iter_copy_from_user_atomic) which is used  by
ext4/tmpfs/pipefs.
Signed-off-by: NTong Tiangen <tongtiangen@huawei.com>

fs/coredump.c

@@ -899,7 +899,9 @@ int dump_user_range(struct coredump_params *cprm, unsigned long start,
 		if (page) {
 			void *kaddr = kmap(page);
 
+			current->flags |= PF_COREDUMP_MCS;
 			stop = !dump_emit(cprm, kaddr, PAGE_SIZE);
+			current->flags &= ~PF_COREDUMP_MCS;
 			kunmap(page);
 			put_page(page);
 		} else {

include/linux/sched.h

@@ -1607,6 +1607,7 @@ extern struct pid *cad_pid;
 #define PF_KTHREAD		0x00200000	/* I am a kernel thread */
 #define PF_RANDOMIZE		0x00400000	/* Randomize virtual address space */
 #define PF_SWAPWRITE		0x00800000	/* Allowed to write to swap */
+#define PF_COREDUMP_MCS		0x01000000      /* Task coredump support machine check safe */
 #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_mask */
 #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
 #define PF_MEMALLOC_NOCMA	0x10000000	/* All allocation request will have _GFP_MOVABLE cleared */

lib/iov_iter.c

@@ -764,6 +764,14 @@ size_t _copy_mc_to_iter(const void *addr, size_t bytes, struct iov_iter *i)
 EXPORT_SYMBOL_GPL(_copy_mc_to_iter);
 #endif /* CONFIG_ARCH_HAS_COPY_MC */
 
+static void *memcpy_iter(void *to, const void *from, __kernel_size_t size)
+{
+	if (IS_ENABLED(CONFIG_ARCH_HAS_COPY_MC) && current->flags & PF_COREDUMP_MCS)
+		return (void *)copy_mc_to_kernel(to, from, size);
+	else
+		return memcpy(to, from, size);
+}
+
 size_t _copy_from_iter(void *addr, size_t bytes, struct iov_iter *i)
 {
 	char *to = addr;
@@ -777,7 +785,7 @@ size_t _copy_from_iter(void *addr, size_t bytes, struct iov_iter *i)
 		copyin((to += v.iov_len) - v.iov_len, v.iov_base, v.iov_len),
 		memcpy_from_page((to += v.bv_len) - v.bv_len, v.bv_page,
 				 v.bv_offset, v.bv_len),
-		memcpy((to += v.iov_len) - v.iov_len, v.iov_base, v.iov_len)
+		memcpy_iter((to += v.iov_len) - v.iov_len, v.iov_base, v.iov_len)
 	)
 
 	return bytes;
@@ -1013,7 +1021,7 @@ size_t iov_iter_copy_from_user_atomic(struct page *page,
 		copyin((p += v.iov_len) - v.iov_len, v.iov_base, v.iov_len),
 		memcpy_from_page((p += v.bv_len) - v.bv_len, v.bv_page,
 				 v.bv_offset, v.bv_len),
-		memcpy((p += v.iov_len) - v.iov_len, v.iov_base, v.iov_len)
+		memcpy_iter((p += v.iov_len) - v.iov_len, v.iov_base, v.iov_len)
 	)
 	kunmap_atomic(kaddr);
 	return bytes;