Skip to contentSkip to conversation

s390: Warn if kernel command line contains non-printable EBCDIC characters

PATCH New Ilya Leoshkevich 2 participants 5 messages Sep 2 s390

lore
Share & tools
lore
#0 lore raw

v4 cover s390: Warn if kernel command line contains non-printable EBCDIC characters

v3: https://lore.kernel.org/linux-s390/[email protected]/
v3 -> v4: Switch to ASCII subset of EBCDIC (Sashiko, Heiko).
Add and use isspace_ebc_asc() to satisfy Heiko's comment about newlines in v1.

v2: https://lore.kernel.org/all/[email protected]/
v2 -> v3: Check whether source EBCDIC characters are printable

          (David, Heiko).

v1: https://lore.kernel.org/all/[email protected]/
v1 -> v2: Account for newlines (Heiko).
Minor stylistic improvements: inline int i, bool return type, comment modality.

Hi,

This series adds a warning for unintentional EBCDIC -> ASCII kernel command line translations. Patch 1 adds EBCDIC character classes. Patch 2 uses them for determining whether the command line is printable.

Best regards,
Ilya

Ilya Leoshkevich (2):
s390/ebcdic: Add character classes for the ASCII subset of EBCDIC
s390: Warn if kernel command line contains non-printable EBCDIC
characters

 arch/s390/boot/ipl_parm.c      | 13 ++++++++++-
 arch/s390/include/asm/ebcdic.h |  6 +++++
 arch/s390/kernel/ebcdic.c      | 40 +++++++++++++++++++++++++++++++++-
 3 files changed, 57 insertions(+), 2 deletions(-)
Signature · 1 line
2.55.0
#1 lore raw

PATCH v4 1/2 s390/ebcdic: Add character classes for the ASCII subset of EBCDIC

Add _ctype-like array for EBCDIC characters. Mark only those that also exist in ASCII, loosely based on invariant subset of EBCDIC [1], but with cent, broken bar, not sign, and plus-minus sign, left out.

[1] https://en.wikipedia.org/wiki/EBCDIC#Code_page_layout

---
 arch/s390/include/asm/ebcdic.h |  6 +++++
 arch/s390/kernel/ebcdic.c      | 40 +++++++++++++++++++++++++++++++++-
 2 files changed, 45 insertions(+), 1 deletion(-)
arch/s390/include/asm/ebcdic.h +6 -0
index 7164cb6584355..78d2e37e985d3 100644
@@ -10,6 +10,7 @@
 #ifndef _EBCDIC_H
 #define _EBCDIC_H
 
+#include <linux/ctype.h>
 #include <linux/types.h>
 
 extern __u8 _ascebc_500[256];   /* ASCII -> EBCDIC 500 conversion table */
@@ -18,6 +19,11 @@ extern __u8 _ascebc[256];   /* ASCII -> EBCDIC conversion table */
 extern __u8 _ebcasc[256];   /* EBCDIC -> ASCII conversion table */
 extern __u8 _ebc_tolower[256]; /* EBCDIC -> lowercase */
 extern __u8 _ebc_toupper[256]; /* EBCDIC -> uppercase */
+extern const __u8 _ebctype_asc[256]; /* EBCDIC -> character classes */
+
+#define __ismask_ebc_asc(x)	(_ebctype_asc[(int)(unsigned char)(x)])
+#define isprint_ebc_asc(c)	((__ismask_ebc_asc(c)&(_P|_U|_L|_D|_SP)) != 0)
+#define isspace_ebc_asc(c)	((__ismask_ebc_asc(c)&(_S)) != 0)
 
 static inline void
 codepage_convert(const __u8 *codepage, volatile char *addr, unsigned long nr)
arch/s390/kernel/ebcdic.c +39 -1
index 0e51fa537262b..815b90ef520c0 100644
@@ -392,10 +392,48 @@ __u8 _ebc_toupper[256] =
 	0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF
 };
 
+/*
+ * ASCII subset of EBCDIC -> character classes
+ */
+const __u8 _ebctype_asc[256] = {
+	_C,     _C,    _C,    _C,    _C,    _C|_S, _C,    _C,	/* 0x00-0x07 */
+	_C,     _C,    _C,    _C|_S, _C|_S, _C|_S, _C,    _C,	/* 0x08-0x0f */
+	_C,     _C,    _C,    _C,    _C,    _C|_S, _C,    _C,	/* 0x10-0x17 */
+	_C,     _C,    _C,    _C,    _C,    _C,    _C,    _C,	/* 0x18-0x1f */
+	_C,     _C,    _C,    _C,    _C,    _C|_S, _C,    _C,	/* 0x20-0x27 */
+	_C,     _C,    _C,    _C,    _C,    _C,    _C,    _C,	/* 0x28-0x2f */
+	0,      0,     _C,    _C,    _C,    _C,    _C,    _C,	/* 0x30-0x37 */
+	_C,     _C,    _C,    _C,    _C,    _C,    0,     _C,	/* 0x38-0x3f */
+	_S|_SP, 0,     0,     0,     0,     0,     0,     0,	/* 0x40-0x47 */
+	0,      0,     0,     _P,    _P,    _P,    _P,    _P,	/* 0x48-0x4f */
+	_P,     0,     0,     0,     0,     0,     0,     0,	/* 0x50-0x57 */
+	0,      0,     _P,    _P,    _P,    _P,    _P,    0,	/* 0x58-0x5f */
+	_P,     _P,    0,     0,     0,     0,     0,     0,	/* 0x60-0x67 */
+	0,      0,     0,     _P,    _P,    _P,    _P,    _P,	/* 0x68-0x6f */
+	0,      0,     0,     0,     0,     0,     0,     0,	/* 0x70-0x77 */
+	0,      _P,    _P,    _P,    _P,    _P,    _P,    _P,	/* 0x78-0x7f */
+	0,      _L|_X, _L|_X, _L|_X, _L|_X, _L|_X, _L|_X, _L,	/* 0x80-0x87 */
+	_L,     _L,    0,     0,     0,     0,     0,     0,	/* 0x88-0x8f */
+	0,      _L,    _L,    _L,    _L,    _L,    _L,    _L,	/* 0x90-0x97 */
+	_L,     _L,    0,     0,     0,     0,     0,     0,	/* 0x98-0x9f */
+	0,      _P,    _L,    _L,    _L,    _L,    _L,    _L,	/* 0xa0-0xa7 */
+	_L,     _L,    0,     0,     0,     0,     0,     0,	/* 0xa8-0xaf */
+	_P,     0,     0,     0,     0,     0,     0,     0,	/* 0xb0-0xb7 */
+	0,      0,     _P,    _P,    0,     0,     0,     0,	/* 0xb8-0xbf */
+	_P,     _U|_X, _U|_X, _U|_X, _U|_X, _U|_X, _U|_X, _U,	/* 0xc0-0xc7 */
+	_U,     _U,    0,     0,     0,     0,     0,     0,	/* 0xc8-0xcf */
+	_P,     _U,    _U,    _U,    _U,    _U,    _U,    _U,	/* 0xd0-0xd7 */
+	_U,     _U,    0,     0,     0,     0,     0,     0,	/* 0xd8-0xdf */
+	_P,     0,     _U,    _U,    _U,    _U,    _U,    _U,	/* 0xe0-0xe7 */
+	_U,     _U,    0,     0,     0,     0,     0,     0,	/* 0xe8-0xef */
+	_D,     _D,    _D,    _D,    _D,    _D,    _D,    _D,	/* 0xf0-0xf7 */
+	_D,     _D,    0,     0,     0,     0,     0,     _C,	/* 0xf8-0xff */
+};
+
 EXPORT_SYMBOL(_ascebc_500);
 EXPORT_SYMBOL(_ebcasc_500);
 EXPORT_SYMBOL(_ascebc);
 EXPORT_SYMBOL(_ebcasc);
 EXPORT_SYMBOL(_ebc_tolower);
 EXPORT_SYMBOL(_ebc_toupper);
-
+EXPORT_SYMBOL(_ebctype_asc);
Signature · 1 line
2.55.0
#2 lore raw

Re: [PATCH v4 1/2] s390/ebcdic: Add character classes for the ASCII subset of EBCDIC

Sashiko has reviewed this patch and found no issues. It looks great!
Quoted 7 lines

Add _ctype-like array for EBCDIC characters. Mark only those that also exist in ASCII, loosely based on invariant subset of EBCDIC [1], but with cent, broken bar, not sign, and plus-minus sign, left out.

[1] https://en.wikipedia.org/wiki/EBCDIC#Code_page_layout

Sashiko has reviewed this patch and found no issues. It looks great!

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#3 lore raw

PATCH v4 2/2 s390: Warn if kernel command line contains non-printable EBCDIC characters

In reply to Ilya Leoshkevich

Users may accidentally add multi-byte UTF-8 characters to zipl.conf parmline, for example, by copying snippets containing non-breaking spaces (\xC2\xA0) from web pages.

The kernel will then interpret the entire command line as EBCDIC, making it unusable. Distinguish this situation from the legitimate EBCDIC conversion by looking for non-printable characters and issue a warning.

Accept the following tradeoffs:

* Do not make this a hard failure, there may be a very small number of users who put characters with diacritics on their EBCDIC-encoded command lines.
* Do not use heuristics with arbitrary thresholds, these may fail intermittently on, e.g., punctuation-heavy command lines, and having to tune thresholds in subsequent patches is not desirable.
* Use the ASCII subset of EBCDIC to determine whether characters are
printable: the vast majority of command lines will contain characters only from this subset.

---
 arch/s390/boot/ipl_parm.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
arch/s390/boot/ipl_parm.c +12 -1
index 6bc950b92be76..dc42ea574f8d7 100644
@@ -173,12 +173,23 @@ static inline int has_ebcdic_char(const char *str)
 	return 0;
 }
 
+static inline bool has_nonprintable_ebcdic_char(const char *str)
+{
+	for (; *str; str++)
+		if (!isprint_ebc_asc(*str) && !isspace_ebc_asc(*str))
+			return true;
+	return false;
+}
+
 void setup_boot_command_line(void)
 {
 	parmarea.command_line[COMMAND_LINE_SIZE - 1] = 0;
 	/* convert arch command line to ascii if necessary */
-	if (has_ebcdic_char(parmarea.command_line))
+	if (has_ebcdic_char(parmarea.command_line)) {
+		if (has_nonprintable_ebcdic_char(parmarea.command_line))
+			boot_warn("Kernel command line was treated as EBCDIC, but contains non-printable characters\n");
 		EBCASC(parmarea.command_line, COMMAND_LINE_SIZE);
+	}
 	/* copy arch command line */
 	strscpy(early_command_line, strim(parmarea.command_line));
 
Signature · 1 line
2.55.0
#4 lore raw
Sashiko has reviewed this patch and found no issues. It looks great!
Quoted 21 lines

Users may accidentally add multi-byte UTF-8 characters to zipl.conf parmline, for example, by copying snippets containing non-breaking spaces (\xC2\xA0) from web pages.

The kernel will then interpret the entire command line as EBCDIC, making it unusable. Distinguish this situation from the legitimate EBCDIC conversion by looking for non-printable characters and issue a warning.

Accept the following tradeoffs:

* Do not make this a hard failure, there may be a very small number of users who put characters with diacritics on their EBCDIC-encoded command lines.
* Do not use heuristics with arbitrary thresholds, these may fail intermittently on, e.g., punctuation-heavy command lines, and having to tune thresholds in subsequent patches is not desirable.
* Use the ASCII subset of EBCDIC to determine whether characters are
printable: the vast majority of command lines will contain characters only from this subset.

Quoted 1 line

Sashiko has reviewed this patch and found no issues. It looks great!

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2