Skip to contentSkip to conversation

test/crypto: change cipher offset for esn vector

PATCH Applied Ankur Dwivedi 3 participants 10 messages Jun 5, 2020 – Jul 15, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 test/crypto: change cipher offset for esn vector

In the IPSEC ESN test vector the cipher offset for encryption should be 8 bytes which is the size of esp header. This patch also changes the ciphertext and the digest reference data against which the operaton result is validated.

Fixes 699741912d71 ("test/crypto: add case for auth only trailer")
---
 app/test/test_cryptodev.c                  |   8 +-
 app/test/test_cryptodev_aes_test_vectors.h | 126 ++++++++++++++---------------
 2 files changed, 67 insertions(+), 67 deletions(-)
app/test/test_cryptodev.c +4 -4
index 8f63146..c812f8c 100644
@@ -10204,7 +10204,7 @@ struct test_crypto_vector {
 static const struct test_crypto_vector
 aes128cbc_hmac_sha1_aad_test_vector = {
 	.crypto_algo = RTE_CRYPTO_CIPHER_AES_CBC,
-	.cipher_offset = 12,
+	.cipher_offset = 8,
 	.cipher_len = 496,
 	.cipher_key = {
 		.data = {
@@ -10240,9 +10240,9 @@ struct test_crypto_vector {
 	},
 	.digest = {
 		.data = {
-			0x1F, 0x6A, 0xD2, 0x8B, 0x4B, 0xB3, 0xC0, 0x9E,
-			0x86, 0x9B, 0x3A, 0xF2, 0x00, 0x5B, 0x4F, 0x08,
-			0x62, 0x8D, 0x62, 0x65
+			0x6D, 0xF3, 0x50, 0x79, 0x7A, 0x2A, 0xAC, 0x7F,
+			0xA6, 0xF0, 0xC6, 0x38, 0x1F, 0xA4, 0xDD, 0x9B,
+			0x62, 0x0F, 0xFB, 0x10
 		},
 		.len = 20
 	}
app/test/test_cryptodev_aes_test_vectors.h +63 -63

132 lines not on this page · raw message

Signature · 1 line
1.9.3
#1 lore raw

Hi All,

Can you please review this patch?

Thanks
Ankur

Quoted 188 lines

-----Original Message-----
From: Ankur Dwivedi <[email protected]>
Sent: Friday, June 5, 2020 2:01 PM
To: [email protected]
Cc: [email protected]; [email protected]; [email protected]; [email protected]; Anoob Joseph <[email protected]>; Ankur Dwivedi <[email protected]>
Subject: [PATCH] test/crypto: change cipher offset for esn vector

In the IPSEC ESN test vector the cipher offset for encryption should be 8 bytes which is the size of esp header.
This patch also changes the ciphertext and the digest reference data against which the operaton result is validated.

Fixes 699741912d71 ("test/crypto: add case for auth only trailer")

---
app/test/test_cryptodev.c | 8 +-
app/test/test_cryptodev_aes_test_vectors.h | 126 ++++++++++++++------------- --

2 files changed, 67 insertions(+), 67 deletions(-)
app/test/test_cryptodev.c index renamed from app/test/test_cryptodev.c +0 -0

8f63146..c812f8c 100644

app/test/test_cryptodev.c +4 -4
@@ -10204,7 +10204,7 @@ struct test_crypto_vector {  static const struct
 test_crypto_vector  aes128cbc_hmac_sha1_aad_test_vector = {
 	.crypto_algo = RTE_CRYPTO_CIPHER_AES_CBC,
-	.cipher_offset = 12,
+	.cipher_offset = 8,
 	.cipher_len = 496,
 	.cipher_key = {
 		.data = {
@@ -10240,9 +10240,9 @@ struct test_crypto_vector {
 	},
 	.digest = {
 		.data = {
-			0x1F, 0x6A, 0xD2, 0x8B, 0x4B, 0xB3, 0xC0, 0x9E,
-			0x86, 0x9B, 0x3A, 0xF2, 0x00, 0x5B, 0x4F, 0x08,
-			0x62, 0x8D, 0x62, 0x65
+			0x6D, 0xF3, 0x50, 0x79, 0x7A, 0x2A, 0xAC, 0x7F,
+			0xA6, 0xF0, 0xC6, 0x38, 0x1F, 0xA4, 0xDD, 0x9B,
+			0x62, 0x0F, 0xFB, 0x10
 		},
 		.len = 20
 	}
app/test/test_cryptodev_aes_test_vectors.h +63 -64
index 3c38eee..f56bbdd 100644
@@ -368,69 +368,69 @@
 
 static const uint8_t ciphertext512_aes128cbc_aad[] = {
 	0x57, 0x68, 0x61, 0x74, 0x20, 0x61, 0x20, 0x6C,
-	0x6F, 0x75, 0x73, 0x79, 0x6D, 0x70, 0xB4, 0xAD,
-	0x09, 0x7C, 0xD7, 0x52, 0xD6, 0xF2, 0xBF, 0xD1,
-	0x9D, 0x79, 0xC6, 0xB6, 0x8F, 0x94, 0xEB, 0xD8,
-	0xBA, 0x5E, 0x01, 0x49, 0x7D, 0xB3, 0xC5, 0xFE,
-	0x18, 0xF4, 0xE3, 0x60, 0x8C, 0x84, 0x68, 0x13,
-	0x33, 0x06, 0x85, 0x60, 0xD3, 0xE7, 0x8A, 0xB5,
-	0x23, 0xA2, 0xDE, 0x52, 0x5C, 0xB6, 0x26, 0x37,
-	0xBB, 0x23, 0x8A, 0x38, 0x07, 0x85, 0xB6, 0x2E,
-	0xC3, 0x69, 0x57, 0x79, 0x6B, 0xE4, 0xD7, 0x86,
-	0x23, 0x72, 0x4C, 0x65, 0x49, 0x08, 0x1E, 0xF3,
-	0xCC, 0x71, 0x4C, 0x45, 0x97, 0x03, 0xBC, 0xA0,
-	0x9D, 0xF0, 0x4F, 0x5D, 0xEC, 0x40, 0x6C, 0xC6,
-	0x52, 0xC0, 0x9D, 0x1C, 0xDC, 0x8B, 0xC2, 0xFA,
-	0x35, 0xA7, 0x3A, 0x00, 0x04, 0x1C, 0xA6, 0x91,
-	0x5D, 0xEB, 0x07, 0xA1, 0xB9, 0x3E, 0xD1, 0xB6,
-	0xCA, 0x96, 0xEC, 0x71, 0xF7, 0x7D, 0xB6, 0x09,
-	0x3D, 0x19, 0x6E, 0x75, 0x03, 0xC3, 0x1A, 0x4E,
-	0x5B, 0x4D, 0xEA, 0xD9, 0x92, 0x96, 0x01, 0xFB,
-	0xA3, 0xC2, 0x6D, 0xC4, 0x17, 0x6B, 0xB4, 0x3B,
-	0x1E, 0x87, 0x54, 0x26, 0x95, 0x63, 0x07, 0x73,
-	0xB6, 0xBA, 0x52, 0xD7, 0xA7, 0xD0, 0x9C, 0x75,
-	0x8A, 0xCF, 0xC4, 0x3C, 0x4A, 0x55, 0x0E, 0x53,
-	0xEC, 0xE0, 0x31, 0x51, 0xB7, 0xB7, 0xD2, 0xB4,
-	0xF3, 0x2B, 0x70, 0x6D, 0x15, 0x9E, 0x57, 0x30,
-	0x72, 0xE5, 0xA4, 0x71, 0x5F, 0xA4, 0xE8, 0x7C,
-	0x46, 0x58, 0x36, 0x71, 0x91, 0x55, 0xAA, 0x99,
-	0x3B, 0x3F, 0xF6, 0xA2, 0x9D, 0x27, 0xBF, 0xC2,
-	0x62, 0x2C, 0x85, 0xB7, 0x51, 0xDD, 0xFD, 0x7B,
-	0x8B, 0xB5, 0xDD, 0x2A, 0x73, 0xF8, 0x93, 0x9A,
-	0x3F, 0xAD, 0x1D, 0xF0, 0x46, 0xD1, 0x76, 0x83,
-	0x71, 0x4E, 0xD3, 0x0D, 0x64, 0x8C, 0xC3, 0xE6,
-	0x03, 0xED, 0xE8, 0x53, 0x23, 0x1A, 0xC7, 0x86,
-	0xEB, 0x87, 0xD6, 0x78, 0xF9, 0xFB, 0x9C, 0x1D,
-	0xE7, 0x4E, 0xC0, 0x70, 0x27, 0x7A, 0x43, 0xE2,
-	0x5D, 0xA4, 0x10, 0x40, 0xBE, 0x61, 0x0D, 0x2B,
-	0x25, 0x08, 0x75, 0x91, 0xB5, 0x5A, 0x26, 0xC8,
-	0x32, 0xA7, 0xC6, 0x88, 0xBF, 0x75, 0x94, 0xCC,
-	0x58, 0xA4, 0xFE, 0x2F, 0xF7, 0x5C, 0xD2, 0x36,
-	0x66, 0x55, 0xF0, 0xEA, 0xF5, 0x64, 0x43, 0xE7,
-	0x6D, 0xE0, 0xED, 0xA1, 0x10, 0x0A, 0x84, 0x07,
-	0x11, 0x88, 0xFA, 0xA1, 0xD3, 0xA0, 0x00, 0x5D,
-	0xEB, 0xB5, 0x62, 0x01, 0x72, 0xC1, 0x9B, 0x39,
-	0x0B, 0xD3, 0xAF, 0x04, 0x19, 0x42, 0xEC, 0xFF,
-	0x4B, 0xB3, 0x5E, 0x87, 0x27, 0xE4, 0x26, 0x57,
-	0x76, 0xCD, 0x36, 0x31, 0x5B, 0x94, 0x74, 0xFF,
-	0x33, 0x91, 0xAA, 0xD1, 0x45, 0x34, 0xC2, 0x11,
-	0xF0, 0x35, 0x44, 0xC9, 0xD5, 0xA2, 0x5A, 0xC2,
-	0xE9, 0x9E, 0xCA, 0xE2, 0x6F, 0xD2, 0x40, 0xB4,
-	0x93, 0x42, 0x78, 0x20, 0x92, 0x88, 0xC7, 0x16,
-	0xCF, 0x15, 0x54, 0x7B, 0xE1, 0x46, 0x38, 0x69,
-	0xB8, 0xE4, 0xF1, 0x81, 0xF0, 0x08, 0x6F, 0x92,
-	0x6D, 0x1A, 0xD9, 0x93, 0xFA, 0xD7, 0x35, 0xFE,
-	0x7F, 0x59, 0x43, 0x1D, 0x3A, 0x3B, 0xFC, 0xD0,
-	0x14, 0x95, 0x1E, 0xB2, 0x04, 0x08, 0x4F, 0xC6,
-	0xEA, 0xE8, 0x22, 0xF3, 0xD7, 0x66, 0x93, 0xAA,
-	0xFD, 0xA0, 0xFE, 0x03, 0x96, 0x54, 0x78, 0x35,
-	0x18, 0xED, 0xB7, 0x2F, 0x40, 0xE3, 0x8E, 0x22,
-	0xC6, 0xDA, 0xB0, 0x8E, 0xA0, 0xA1, 0x62, 0x03,
-	0x63, 0x34, 0x11, 0xF5, 0x9E, 0xAA, 0x6B, 0xC4,
-	0x14, 0x75, 0x4C, 0xF4, 0xD8, 0xD9, 0xF1, 0x76,
-	0xE3, 0xD3, 0x55, 0xCE, 0x22, 0x7D, 0x4A, 0xB7,
-	0xBB, 0x7F, 0x4F, 0x09, 0x88, 0x70, 0x6E, 0x09,
-	0x84, 0x6B, 0x24, 0x19, 0x2C, 0x20, 0x73, 0x75
+	0x1D, 0x7C, 0x76, 0xED, 0xC2, 0x10, 0x3C, 0xB5,
+	0x14, 0x07, 0x3C, 0x33, 0x7B, 0xBE, 0x9E, 0xA9,
+	0x01, 0xC5, 0xAA, 0xA6, 0xB6, 0x7A, 0xE1, 0xDB,
+	0x39, 0xAA, 0xAA, 0xF4, 0xEE, 0xA7, 0x71, 0x71,
+	0x78, 0x0D, 0x5A, 0xD4, 0xF9, 0xCD, 0x75, 0xD1,
+	0x9C, 0x7F, 0xC8, 0x58, 0x46, 0x7A, 0xD1, 0x81,
+	0xEA, 0xCC, 0x08, 0xDC, 0x82, 0x73, 0x22, 0x08,
+	0x11, 0x73, 0x7C, 0xB1, 0x84, 0x6A, 0x8E, 0x67,
+	0x3F, 0x5D, 0xDB, 0x0E, 0xE2, 0xC2, 0xCB, 0x6D,
+	0x88, 0xEC, 0x3F, 0x50, 0x44, 0xD3, 0x47, 0x6E,
+	0xDD, 0x42, 0xDC, 0x2A, 0x5E, 0x5C, 0x50, 0x24,
+	0x57, 0x8A, 0xE7, 0xC5, 0x53, 0x6D, 0x89, 0x33,
+	0x21, 0x65, 0x82, 0xD6, 0xE9, 0xE7, 0x77, 0x10,
+	0xC2, 0x09, 0x91, 0xC1, 0x42, 0x62, 0x36, 0xF4,
+	0x43, 0x37, 0x95, 0xB3, 0x7E, 0x21, 0xC5, 0x3E,
+	0x65, 0xCB, 0xB6, 0xAA, 0xEC, 0xA5, 0xC6, 0x5C,
+	0x4D, 0xBE, 0x14, 0xF1, 0x98, 0xBF, 0x6C, 0x8A,
+	0x9E, 0x9F, 0xD4, 0xB4, 0xF2, 0x22, 0x96, 0x99,
+	0x37, 0x32, 0xB6, 0xC1, 0x04, 0x66, 0x52, 0x37,
+	0x5D, 0x5F, 0x58, 0x92, 0xC9, 0x97, 0xEA, 0x60,
+	0x60, 0x27, 0x57, 0xF9, 0x47, 0x4F, 0xBC, 0xDF,
+	0x05, 0xBD, 0x37, 0x87, 0xBB, 0x09, 0xA5, 0xBE,
+	0xC1, 0xFC, 0x32, 0x86, 0x6A, 0xB7, 0x8B, 0x1E,
+	0x6B, 0xCE, 0x8D, 0x81, 0x63, 0x4C, 0xF2, 0x7F,
+	0xD1, 0x45, 0x82, 0xE8, 0x0D, 0x1C, 0x4D, 0xA8,
+	0xBF, 0x2D, 0x2B, 0x52, 0xE5, 0xDB, 0xAB, 0xFD,
+	0x04, 0xA2, 0xA1, 0x1E, 0x21, 0x1D, 0x06, 0x9A,
+	0xC2, 0x7D, 0x99, 0xFC, 0xB4, 0x72, 0x89, 0x41,
+	0x55, 0x69, 0xFA, 0x1F, 0x78, 0x2F, 0x35, 0x59,
+	0xD7, 0x59, 0x6D, 0xA6, 0x45, 0xC9, 0x2B, 0x06,
+	0x6C, 0xEC, 0x83, 0x34, 0xA5, 0x08, 0xDB, 0x6F,
+	0xDE, 0x75, 0x21, 0x9B, 0xB0, 0xCB, 0x0A, 0xAE,
+	0x22, 0x99, 0x74, 0x1C, 0x9D, 0x37, 0x0E, 0xC6,
+	0x3A, 0x45, 0x49, 0xE5, 0xE3, 0x21, 0x11, 0xEA,
+	0x34, 0x25, 0xD5, 0x76, 0xB0, 0x30, 0x19, 0x87,
+	0x14, 0x3A, 0x10, 0x6F, 0x6D, 0xDD, 0xE9, 0x60,
+	0x6A, 0x00, 0x6A, 0x4C, 0x5B, 0x85, 0x3E, 0x1A,
+	0x41, 0xFA, 0xDE, 0x2D, 0x2F, 0x2E, 0x5B, 0x79,
+	0x09, 0x66, 0x65, 0xD0, 0xDB, 0x32, 0x05, 0xB5,
+	0xEA, 0xFB, 0x6A, 0xD5, 0x43, 0xF8, 0xBD, 0x98,
+	0x7B, 0x8E, 0x3B, 0x85, 0x89, 0x5D, 0xC5, 0x59,
+	0x54, 0x22, 0x75, 0xA8, 0x60, 0xDC, 0x0A, 0x37,
+	0x8C, 0xD8, 0x05, 0xEA, 0x62, 0x62, 0x71, 0x98,
+	0x0C, 0xCB, 0xCE, 0x0A, 0xD9, 0xE6, 0xE8, 0xA7,
+	0xB3, 0x2D, 0x89, 0xA7, 0x60, 0xF0, 0x42, 0xA7,
+	0x3D, 0x80, 0x44, 0xE7, 0xC1, 0xA6, 0x88, 0xB1,
+	0x4F, 0xC0, 0xB1, 0xAF, 0x40, 0xF3, 0x54, 0x72,
+	0x8F, 0xAF, 0x47, 0x96, 0x19, 0xEB, 0xA5, 0x5C,
+	0x00, 0x3B, 0x36, 0xC8, 0x3F, 0x1E, 0x63, 0x54,
+	0xF3, 0x3D, 0x85, 0x44, 0x9B, 0x9B, 0x20, 0xE3,
+	0x9D, 0xEF, 0x62, 0x21, 0xA1, 0x0B, 0x78, 0xF4,
+	0x2B, 0x89, 0x66, 0x5E, 0x97, 0xC6, 0xC4, 0x55,
+	0x35, 0x32, 0xD7, 0x44, 0x95, 0x9A, 0xE7, 0xF2,
+	0x57, 0x52, 0x5B, 0x92, 0x86, 0x8F, 0x8B, 0xCF,
+	0x41, 0x89, 0xF6, 0x2A, 0xD3, 0x42, 0x87, 0x43,
+	0x56, 0x1F, 0x0E, 0x49, 0xF1, 0x32, 0x6D, 0xA8,
+	0x62, 0xDF, 0x47, 0xBB, 0xB6, 0x53, 0xF8, 0x5C,
+	0x36, 0xDA, 0x34, 0x34, 0x2D, 0x2E, 0x1D, 0x33,
+	0xAF, 0x6A, 0x1E, 0xF1, 0xC9, 0x72, 0xB5, 0x3C,
+	0x64, 0x4C, 0x96, 0x12, 0x78, 0x67, 0x6A, 0xE5,
+	0x8B, 0x05, 0x80, 0xAE, 0x7D, 0xE5, 0x9B, 0x24,
+	0xDB, 0xFF, 0x1E, 0xB8, 0x36, 0x6D, 0x3D, 0x5D,
+	0x73, 0x65, 0x72, 0x73, 0x2C, 0x20, 0x73, 0x75
 };
 
 /* AES128-CTR-SHA1 test vector */
--

1.9.3

#2 lore raw
In reply to Ankur Dwivedi

Hi Ankur,

Quoted 10 lines

In the IPSEC ESN test vector the cipher offset for encryption should be 8 bytes which is the size of esp header. This patch also changes the ciphertext and the digest reference data against which the operaton result is validated.

Fixes 699741912d71 ("test/crypto: add case for auth only trailer")

---

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app requirement.
I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

Regards,
Akhil

#3 lore raw

Hi Akhil,

Please see my reply inline.

Thanks
Ankur

Quoted 24 lines

-----Original Message-----
From: Akhil Goyal <[email protected]>
Sent: Monday, July 6, 2020 5:54 PM
To: Ankur Dwivedi <[email protected]>; [email protected]
Cc: [email protected]; Hemant Agrawal <[email protected]>; Vakul Garg <[email protected]>; Anoob Joseph <[email protected]>
Subject: [EXT] RE: [PATCH] test/crypto: change cipher offset for esn vector

External Email

----------------------------------------------------------------------
Hi Ankur,

Quoted 10 lines

In the IPSEC ESN test vector the cipher offset for encryption should be 8 bytes which is the size of esp header.
This patch also changes the ciphertext and the digest reference data against which the operaton result is validated.

Fixes 699741912d71 ("test/crypto: add case for auth only trailer")

---

Quoted 6 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app requirement.
I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned. In the IPSEC ESN scenario generally the offsets will be such.
But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

Quoted 3 lines

Regards,
Akhil

#4 lore raw
Quoted 11 lines
Quoted 6 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app requirement.
I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such.
But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

#5 lore raw

Hi Akhil,

Please see my reply inline.

Thanks
Ankur

Quoted 25 lines

-----Original Message-----
From: Akhil Goyal <[email protected]>
Sent: Tuesday, July 7, 2020 12:24 AM
To: Ankur Dwivedi <[email protected]>; [email protected]
Cc: [email protected]; Hemant Agrawal <[email protected]>; Vakul Garg <[email protected]>; Anoob Joseph <[email protected]>
Subject: [EXT] RE: [PATCH] test/crypto: change cipher offset for esn vector

External Email

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

Quoted 4 lines
Quoted 4 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app

requirement.

Quoted 7 lines
Quoted 2 lines

I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such. But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

Quoted 3 lines

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

[Ankur] The offsets values are present per crypto operation. So to make these tests as unsupported the pmd datapath needs to be changed.
Is there an alternative to make these tests unsupported?

#6 lore raw
Quoted 14 lines
Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app

requirement.

Quoted 7 lines
Quoted 2 lines

I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such. But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

Quoted 7 lines
Quoted 3 lines

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

[Ankur] The offsets values are present per crypto operation. So to make these tests
as unsupported the pmd datapath needs to be changed.
Is there an alternative to make these tests unsupported?

I believe this is a data path error and a limitation in your PMD.
You can not stop the application writer from using unaligned cipher offsets(non-8 byte aligned)

This is just a test application, which may hide your PMD limitation by accepting this patch But in actual the scenario will fail when some user configures a 12B cipher offset. IMO, you should add a new test vector instead of replacing this one and it should Be OK to have the existing one fail in your case.

#7 lore raw

Hi Akhil,

Please see inline.

Thanks,
Anoob

Quoted 30 lines

-----Original Message-----
From: Akhil Goyal <[email protected]>
Sent: Wednesday, July 8, 2020 12:51 AM
To: Ankur Dwivedi <[email protected]>; [email protected]
Cc: [email protected]; Hemant Agrawal <[email protected]>; Vakul Garg <[email protected]>; Anoob Joseph <[email protected]>
Subject: [EXT] RE: [PATCH] test/crypto: change cipher offset for esn vector

External Email

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

Quoted 17 lines
Quoted 17 lines
Quoted 5 lines
Quoted 5 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app

requirement.

Quoted 7 lines
Quoted 2 lines

I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such. But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

Quoted 7 lines
Quoted 3 lines

[Ankur] The offsets values are present per crypto operation. So to make these tests as unsupported the pmd datapath needs to be changed. Is there an alternative to make these tests unsupported?

I believe this is a data path error and a limitation in your PMD. You can not stop the application writer from using unaligned cipher offsets(non-8 byte aligned)

[Anoob] Yes. But the typical case with IPsec is what is supported in the PMD.

Quoted 4 lines

This is just a test application, which may hide your PMD limitation by accepting this patch But in actual the scenario will fail when some user configures a 12B cipher offset.

[Anoob] Agreed. But autotest having a failure is not an ideal situation to be it. Especially when it's not the typical usage. Can I propose to add a field like "Know Issues:" in the summary field? We can add a check for OCTEONTX PMDs in the test case and mark it as a known case. I do understand that the vision was to remove all driver specific tests and have generic tests for all PMDs, but here we are left with no other option. Chances are, other PMDs also could have similar limitations when moving to generic framework.

If you have suggestions to skip this test in any other way, that would also work for us.

Quoted 3 lines

IMO, you should add a new test vector instead of replacing this one and it should Be OK to have the existing one fail in your case.

#8 lore raw

Hi Akhil,

Did you get a chance to look at my comments?

Thanks,
Anoob

Quoted 78 lines

-----Original Message-----
From: dev <[email protected]> On Behalf Of Anoob Joseph
Sent: Wednesday, July 8, 2020 3:21 PM
To: Akhil Goyal <[email protected]>; Ankur Dwivedi <[email protected]>; [email protected]
Cc: [email protected]; Hemant Agrawal <[email protected]>; Vakul Garg <[email protected]>
Subject: Re: [dpdk-dev] [PATCH] test/crypto: change cipher offset for esn vector

Hi Akhil,

Please see inline.

Thanks,
Anoob

Quoted 5 lines

-----Original Message-----
From: Akhil Goyal <[email protected]>
Sent: Wednesday, July 8, 2020 12:51 AM
To: Ankur Dwivedi <[email protected]>; [email protected]
Cc: [email protected]; Hemant Agrawal

<[email protected]>;

Quoted 4 lines

Vakul Garg <[email protected]>; Anoob Joseph <[email protected]>
Subject: [EXT] RE: [PATCH] test/crypto: change cipher offset for esn vector

Quoted 23 lines

External Email

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

Quoted 19 lines
Quoted 14 lines
Quoted 5 lines
Quoted 5 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app

requirement.

Quoted 7 lines
Quoted 2 lines

I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset
- auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such. But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

Quoted 3 lines

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

[Ankur] The offsets values are present per crypto operation. So to make these tests as unsupported the pmd datapath needs to be

changed.

Quoted 5 lines
Quoted 1 line

Is there an alternative to make these tests unsupported?

I believe this is a data path error and a limitation in your PMD. You can not stop the application writer from using unaligned cipher offsets(non-8 byte aligned)

[Anoob] Yes. But the typical case with IPsec is what is supported in the PMD.

Quoted 4 lines

This is just a test application, which may hide your PMD limitation by accepting this patch But in actual the scenario will fail when some user configures a 12B cipher offset.

[Anoob] Agreed. But autotest having a failure is not an ideal situation to be it. Especially when it's not the typical usage. Can I propose to add a field like "Know Issues:" in the summary field? We can add a check for OCTEONTX PMDs in the test case and mark it as a known case. I do understand that the vision was to remove all driver specific tests and have generic tests for all PMDs, but here we are left with no other option. Chances are, other PMDs also could have similar limitations when moving to generic framework.

If you have suggestions to skip this test in any other way, that would also work for us.

Quoted 3 lines

IMO, you should add a new test vector instead of replacing this one and it should Be OK to have the existing one fail in your case.

Applied
#9 lore raw
In reply to Anoob Joseph
Quoted 41 lines
Quoted 18 lines
Quoted 18 lines
Quoted 18 lines
Quoted 6 lines
Quoted 6 lines

Why do we need this change?
The existing test case is to demonstrate a generic case where we can have an auth only trailer as well. It is similar to a case of IPSEC ESN but not exactly IPSec. Cipher offset can be anything as per the app

requirement.

Quoted 7 lines
Quoted 2 lines

I don't think there is anything wrong in the vector. It should pass in every hardware without any issue.

[Ankur] It's a limitation in OCTEON TX PMDs that the (encr offset - auth offset) should be 8 bytes aligned.
In the IPSEC ESN scenario generally the offsets will be such. But in the above IPSEC ESN test vector, this requirement is not met and hence the associated test cases fails on the PMD.

In that case, I think it is better to have a separate test vector and both should be executed. With the previous one as not supported in your case and this one will be supported.

Quoted 7 lines
Quoted 3 lines

[Ankur] The offsets values are present per crypto operation. So to make these tests as unsupported the pmd datapath needs to be changed. Is there an alternative to make these tests unsupported?

I believe this is a data path error and a limitation in your PMD. You can not stop the application writer from using unaligned cipher offsets(non-8 byte aligned)

[Anoob] Yes. But the typical case with IPsec is what is supported in the PMD.

Quoted 4 lines

This is just a test application, which may hide your PMD limitation by accepting this patch But in actual the scenario will fail when some user configures a 12B cipher offset.

[Anoob] Agreed. But autotest having a failure is not an ideal situation to be it. Especially when it's not the typical usage. Can I propose to add a field like "Know Issues:" in the summary field? We can add a check for OCTEONTX PMDs in the test case and mark it as a known case. I do understand that the vision was to remove all driver specific tests and have generic tests for all PMDs, but here we are left with no other option. Chances are, other PMDs also could have similar limitations when moving to generic framework.

Quoted 3 lines

If you have suggestions to skip this test in any other way, that would also work for us.

I don't see there is a straight forward way for skipping this test, but we cannot allow PMD specific checks going further.
If you are not ready to accept this failure case, I am ok to take this patch as is. I believe it will hide your PMD limitation and may come up
As a surprise to your customers if they start using non-8 byte aligned cipher offset.

Applied to dpdk-next-crypto

Thanks.

Quoted 3 lines
Quoted 3 lines

IMO, you should add a new test vector instead of replacing this one and it should Be OK to have the existing one fail in your case.