Wipe excluded pages that are written into ELF dump file

When a page is excluded by any of the existing dump levels, that page may still be written to the ELF dump file, depending upon the PFN_EXCLUDED mechanism.

The PFN_EXCLUDED mechanism looks for N consecutive "not dumpable" pages, and if found, the current ELF segment is closed out and a new ELF segment started, at the next dumpable page. Otherwise, if the PFN_EXCLUDED criteria is not meet (that is, there is a mix of dumpable and not dumpable pages, but not N consecutive not dumpable pages) all pages are written to the dump file.

This patch implements a mechanism for those "not dumpable" pages that are written to the ELF dump file to fill those pages with constant data, rather than the original data. In other words, the dump file still contains the page, but its data is wiped. The data is wiped with the value 0xDEAD9A6EDEAD9A6EUL (an attempt at DEADPAGE in hex, which works for 32-bit targets as well).

The motivation for doing this is to protect real user (customer) data from "leaking" through to a dump file when that data was intended to be omitted.

Signed-off-by Eric DeVolder <[email protected]>

---
v2: posted 04aug2017 to mailing list
- Incorporate feedback from Daniel Kiper (wipe value)
- Incorporate feedback from Atsushi Kumagai (eliminate the option and make as default/builtin behavior)

v1: posted 31jul2017 to mailing list
---
makedumpfile.c | 27 ++++++++++++++++++++-------
makedumpfile.h | 1 +

2 files changed, 21 insertions(+), 7 deletions(-)

makedumpfile.c +0 -0

index f85003a..66c3105 100644
@@ -7139,7 +7139,7 @@ out:
 

int
write_elf_load_segment(struct cache_data *cd_page, unsigned long long paddr,

makedumpfile.c +1 -1

-		       off_t off_memory, long long size)
+		       off_t off_memory, long long size, struct cycle *cycle)

{
	long page_size = info->page_size;
	long long bufsz_write;

makedumpfile.c +19 -6

@@ -7163,10 +7163,23 @@ write_elf_load_segment(struct cache_data *cd_page, unsigned long long paddr,
 		else
 			bufsz_write = size;
 
-		if (read(info->fd_memory, buf, bufsz_write) != bufsz_write) {
-			ERRMSG("Can't read the dump memory(%s). %s\n",
-			    info->name_memory, strerror(errno));
-			return FALSE;
+		if (!is_dumpable(info->bitmap2, paddr_to_pfn(paddr), cycle)) {
+			unsigned k;
+			unsigned long *p = (unsigned long *)buf;
+			for (k = 0; k < info->page_size; k += sizeof(unsigned long)) {
+				*p++ = FILL_EXCLUDED_PAGES_VALUE;
+			}
+			if (lseek(info->fd_memory, bufsz_write, SEEK_CUR) < 0) {
+				ERRMSG("Can't seek the dump memory(%s). %s\n",
+				    info->name_memory, strerror(errno));
+				return FALSE;
+			}
+		} else {
+			if (read(info->fd_memory, buf, bufsz_write) != bufsz_write) {
+				ERRMSG("Can't read the dump memory(%s). %s\n",
+				    info->name_memory, strerror(errno));
+				return FALSE;
+			}
 		}
 		filter_data_buffer((unsigned char *)buf, paddr, bufsz_write);
 		paddr += bufsz_write;

@@ -7431,7 +7444,7 @@ write_elf_pages_cyclic(struct cache_data *cd_header, struct cache_data *cd_page)
 				 */
 				if (load.p_filesz)
 					if (!write_elf_load_segment(cd_page, paddr,
-								    off_memory, load.p_filesz))
+								    off_memory, load.p_filesz, &cycle))
 						return FALSE;
 
 				load.p_paddr += load.p_memsz;

@@ -7473,7 +7486,7 @@ write_elf_pages_cyclic(struct cache_data *cd_header, struct cache_data *cd_page)
 		 */
 		if (load.p_filesz)
 			if (!write_elf_load_segment(cd_page, paddr,
-						    off_memory, load.p_filesz))
+						    off_memory, load.p_filesz, &cycle))
 				return FALSE;
 
 		off_seg_load += load.p_filesz;

makedumpfile.h +1 -1

index 8a05794..e043cf2 100644
@@ -223,6 +223,7 @@ isAnon(unsigned long mapping)
 #define FILENAME_BITMAP		"kdump_bitmapXXXXXX"
 #define FILENAME_STDOUT		"STDOUT"
 #define MAP_REGION		(4096*1024)
+#define FILL_EXCLUDED_PAGES_VALUE   (0xDEAD9A6EDEAD9A6EUL)
 
 /*
 * Minimam vmcore has 2 ProgramHeaderTables(PT_NOTE and PT_LOAD).
--

2.7.4

Quoted 7 lines

Hello Eric,

Thanks for your work, this version looks good to me. I'll merge this into v1.6.3.

Regards,
Atsushi Kumagai

Thank you!
eric

Hello Eric,

Quoted 11 lines · Atsushi Kumagai

Quoted 7 lines

Hello Eric,

Thanks for your work, this version looks good to me. I'll merge this into v1.6.3.

Regards,
Atsushi Kumagai

Thank you!
eric

I noticed that wiping zero pages seems to cause different behavior of crash like:

[kmem -i]

(quoted diff) +2 -2

-    SWAP USED        1         4 KB    0% of TOTAL SWAP
-    SWAP FREE   489979       1.9 GB   99% of TOTAL SWAP
+    SWAP USED   229886       898 MB   46% of TOTAL SWAP  // wiped page is counted as SWAP USED ?
+    SWAP FREE   260094      1016 MB   53% of TOTAL SWAP

I don't investigate the details yet, but I decided to slip this feature to v1.6.4.

Thanks,
Atsushi Kumagai

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec