mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] fs/udf/ecma_167.h hex offsets
@ 2003-12-17 21:09 Pat LaVarre
  0 siblings, 0 replies; only message in thread
From: Pat LaVarre @ 2003-12-17 21:09 UTC (permalink / raw)
  To: linux-kernel

I see ecma-167.pdf publishes only the useless decimal offsets, not the
plain hex I see in bus traces, core dumps, disc images, etc.

fs/udf/ecma_167.h as yet has neither kind of offset.  Courtesy `dc -e
'216 212 208 200 168 152 136 132 128 116 104 92 80 72 64 56 52 51 50 48
44 40 36 16 0 16dio f' | less` we could choose to comment.

Pat LaVarre

--- linux-2.6.0-test11/fs/udf/ecma_167.h	2003-11-26 13:43:39.000000000
-0700
+++ linux/fs/udf/ecma_167.h	2003-12-17 14:05:12.217784376 -0700
@@ -791,32 +791,32 @@ struct pathComponent
 /* File Entry (ECMA 167r3 4/14.17) */
 struct extendedFileEntry
 {
-	tag		descTag;
-	icbtag		icbTag;
-	uint32_t	uid;
-	uint32_t	gid;
-	uint32_t	permissions;
-	uint16_t	fileLinkCount;
-	uint8_t		recordFormat;
-	uint8_t		recordDisplayAttr;
-	uint32_t	recordLength;
-	uint64_t	informationLength;
-	uint64_t	objectSize;
-	uint64_t	logicalBlocksRecorded;
-	timestamp	accessTime;
-	timestamp	modificationTime;
-	timestamp	createTime;
-	timestamp	attrTime;
-	uint32_t	checkpoint;
-	uint32_t	reserved;
-	long_ad		extendedAttrICB;
-	long_ad		streamDirectoryICB;
-	regid		impIdent;
-	uint64_t	uniqueID;
-	uint32_t	lengthExtendedAttr;
-	uint32_t	lengthAllocDescs;
-	uint8_t		extendedAttr[0];
-	uint8_t		allocDescs[0];
+	tag		descTag;		/* 0 */
+	icbtag		icbTag;			/* 16=x10 */
+	uint32_t	uid;			/* 36=x24 */
+	uint32_t	gid;			/* 40=x28 */
+	uint32_t	permissions;		/* 44=x2C */
+	uint16_t	fileLinkCount;		/* 48=x30 */
+	uint8_t		recordFormat;		/* 50=x32 */
+	uint8_t		recordDisplayAttr;	/* 51=x33 */
+	uint32_t	recordLength;		/* 52=x34 */
+	uint64_t	informationLength;	/* 56=x38 */
+	uint64_t	objectSize;		/* 64=x40 */
+	uint64_t	logicalBlocksRecorded;	/* 72=x48 */
+	timestamp	accessTime;		/* 80=x50 */
+	timestamp	modificationTime;	/* 92=x5C */
+	timestamp	createTime;		/* 104=x68 */
+	timestamp	attrTime;		/* 116=x74 */
+	uint32_t	checkpoint;		/* 128=x80 */
+	uint32_t	reserved;		/* 132=x84 */
+	long_ad		extendedAttrICB;	/* 136=x88 */
+	long_ad		streamDirectoryICB;	/* 152=x98 */
+	regid		impIdent;		/* 168=xA8 */
+	uint64_t	uniqueID;		/* 200=xC8 */
+	uint32_t	lengthExtendedAttr;	/* 208=xD0 */
+	uint32_t	lengthAllocDescs;	/* 212=xD4 */
+	uint8_t		extendedAttr[0];	/* 216=xD8 + L_EA */
+	uint8_t		allocDescs[0];		/* 216=xD8 + L_EA */
 } __attribute__ ((packed));
 
 #endif /* _ECMA_167_H */



^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2003-12-17 21:09 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-12-17 21:09 [PATCH] fs/udf/ecma_167.h hex offsets Pat LaVarre

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®