| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <eb2a9138-a76e-4b1e-e001-315149c4dabd@leica-geosystems.com>
Date: Fri, 6 Oct 2023 10:30:36 +0000
From: SHUKLA Mamta Ramendra <mamta.shukla@...ca-geosystems.com>
To: Tudor Ambarus <tudor.ambarus@...aro.org>,
"pratyush@...nel.org" <pratyush@...nel.org>,
"michael@...le.cc" <michael@...le.cc>,
"miquel.raynal@...tlin.com" <miquel.raynal@...tlin.com>,
"richard@....at" <richard@....at>,
"vigneshr@...com" <vigneshr@...com>,
"linux-mtd@...ts.infradead.org" <linux-mtd@...ts.infradead.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
CC: GEO-CHHER-bsp-development
<bsp-development.geo@...ca-geosystems.com>
Subject: Re: [PATCH v2] mtd: micron-st: enable lock/unlock for mt25qu512a
>>> On 10/5/23 10:21, SHUKLA Mamta Ramendra wrote:
>>>
>>> cut
>>>
>>>>>
>>>>> and the second one will add just the BP support, something like:
>>>>> diff --git a/drivers/mtd/spi-nor/micron-st.c
>>>>> b/drivers/mtd/spi-nor/micron-st.c
>>>>> index a8da1f18e335..fdafbfa0f936 100644
>>>>> --- a/drivers/mtd/spi-nor/micron-st.c
>>>>> +++ b/drivers/mtd/spi-nor/micron-st.c
>>>>> @@ -405,6 +405,8 @@ static const struct flash_info st_nor_parts[] = {
>>>>> }, {
>>>>> .id = SNOR_ID(0x20, 0xbb, 0x20, 0x10, 0x44, 0x00),
>>>>> .name = "mt25qu512a",
>>>>> + .flags = SPI_NOR_HAS_LOCK | SPI_NOR_HAS_TB |
>>>>> SPI_NOR_4BIT_BP |
>>>>> + SPI_NOR_BP3_SR_BIT6,
>>>>> .mfr_flags = USE_FSR,
>>>>> }, {
>>>>> .id = SNOR_ID(0x20, 0xbb, 0x20),
>>>>>
>>>>> Of course, I expect you to run again the mtd_debug tests and also verify
>>>>> the locking. Thanks!
>>>>
>>>> I applied both changes as mentioned above i.e
>>>> 1] Switch to SFDP and 2] Using BP Flags.
>>>>
>>>> Case 1: Use BP Flags and Switch to SFDP
>>>> With both these changes, the lock/unlock doesn't work.
>>>>
>>>> ## x86-64-smarc-evk-uwd0j0007 # uname -r
>>>> 6.6.0-rc2
>>>>
>>>> # flash_lock -i /dev/mtd0
>>>> Device: /dev/mtd0
>>>> Start: 0
>>>> Len: 0x4000000
>>>> Lock status: unlocked
>>>> Return code: 0
>>>> # flash_lock -l /dev/mtd0
>>>> flash_lock: error!: could not lock device: /dev/mtd0
>>>>
>>>> error 5 (Input/output error)
>>>>
>>>>
>>>> I suspected this is because of miscalculation of BP bits, like the
>>>> possibility mentioned here:
>>>> https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git/commit/?h=spi-nor/next&id=3ea3f0ac242c86c0275d347ab8c92bf1eb854b49
>>>>
>>>>
>>>> But further checked size, it is correct:
>>>>
>>>> # mtd_debug info /dev/mtd0
>>>> mtd.type = MTD_NORFLASH
>>>> mtd.flags = MTD_CAP_NORFLASH
>>>> mtd.size = 67108864 (64M)
>>>> mtd.erasesize = 4096 (4K)
>>>> mtd.writesize = 1
>>>> mtd.oobsize = 0
>>>> regions = 0
>>>>
>>>> And rest of read/write functions work as expected.
>>>>
>>>> Any suggestions about this?
>>>>
>>>> Case 2: Just added BP flags, rest of the size, mfr_flags, fixup flags
>>>> kept as it is.
>>>
>>> would you please detail what exact definitions you used in case 2? Send
>>> us the diff please.
>> Case 2: Adding Flags for BP
>>
>> diff --git a/drivers/mtd/spi-nor/micron-st.c
>> b/drivers/mtd/spi-nor/micron-st.c
>> index 4afcfc57c896..6c8cabbead2e 100644
>> --- a/drivers/mtd/spi-nor/micron-st.c
>> +++ b/drivers/mtd/spi-nor/micron-st.c
>> @@ -405,6 +405,8 @@ static const struct flash_info st_nor_parts[] = {
>> }, {
>> .id = SNOR_ID(0x20, 0xbb, 0x20, 0x10, 0x44, 0x00),
>> .name = "mt25qu512a",
>> + .flags = SPI_NOR_HAS_LOCK | SPI_NOR_HAS_TB |
>> SPI_NOR_4BIT_BP |
>> + SPI_NOR_BP3_SR_BIT6,
>> .size = SZ_64M,
>> .no_sfdp_flags = SECT_4K | SPI_NOR_DUAL_READ |
>> SPI_NOR_QUAD_READ,
>> .fixup_flags = SPI_NOR_4B_OPCODES,
>>
>>
>
> Thanks. Nothing obvious on a first look. I looked at the sfdp dump, it
> looks like 4BAIT table is missing, so you'll probably need:
> .fixup_flags = SPI_NOR_4B_OPCODES,
> I don't see how this could affect BP, but it is worth to test
> incremental changes and find out what misses.
>
> After you test the above, I'd like you to extend the patch with:
> .size = SZ_64M,
> Check if it works and send us the output of mtd_debug info /dev/mtd0 here.
I started first with params and debug log for working and non-working cases:
1] Non-working case: BP Flags + Parsing SFDP(i.e remove size, fixup flag
and no_sfdp flag)
## x86-64-smarc-evk-uwd0j0007 # flash_lock -l /dev/mtd0
[ 480.230232] spi-nor spi-PRP0001:00: SR: Read back test failed
flash_lock: error!: could not lock device: /dev/mtd0
error 5 (Input/output error)
and params from debugfs look like:
## x86-64-smarc-evk-uwd0j0007 # cat
/sys/kernel/debug/spi-nor/spi-PRP0001:00/params
name mt25qu512a
id 20 bb 20 10 44 00
size 64.0 MiB
write size 1
page size 256
address nbytes 4
flags HAS_SR_TB | 4B_OPCODES | HAS_4BAIT | HAS_LOCK |
HAS_16BIT_SR | HAS_4BIT_BP | HAS_SR_BP3_BIT6 | SOFT_RESET
Strange thing here is HAS_16BIT_SR. And it makes sense from debug log:
[ 480.230232] spi-nor spi-PRP0001:00: SR: Read back test failed
which comes from spi_nor_write_16bit_sr_and_check () .
It is strange because in micron-st.c, in default init we have:
static void micron_st_nor_default_init(struct spi_nor *nor)
{
nor->flags |= SNOR_F_HAS_LOCK;
nor->flags &= ~SNOR_F_HAS_16BIT_SR;
nor->params->quad_enable = NULL;
}
2] Working case: Just BP flags and no parsing SFDP:
Params for flags look like
flags HAS_SR_TB | 4B_OPCODES | HAS_4BAIT | HAS_LOCK |
HAS_4BIT_BP | HAS_SR_BP3_BIT6 | SOFT_RESET
3] Just adding .fixup_flags:
@@ -407,6 +418,7 @@ static const struct flash_info st_nor_parts[] = {
.name = "mt25qu512a",
.flags = SPI_NOR_HAS_LOCK | SPI_NOR_HAS_TB |
SPI_NOR_4BIT_BP |
SPI_NOR_BP3_SR_BIT6,
+ .fixup_flags = SPI_NOR_4B_OPCODES,
.mfr_flags = USE_FSR,
lock/unlock still doesn't work.
I also compared this with case when we just parse SFDP flag, we get
4Byte-AIT, so this flag is not required when parsing SFDP.
This is only required when we don't parse SFDP and use .size=SZ_64M
4] Adding .size flag i.e no SFDP parsing:
lock/unlock works as expected and params:
## x86-64-smarc-evk-uwd0j0007 # cat
/sys/kernel/debug/spi-nor/spi-PRP0001:00/params
name mt25qu512a
id 20 bb 20 10 44 00
size 64.0 MiB
write size 1
page size 256
address nbytes 4
flags HAS_SR_TB | 4B_OPCODES | HAS_LOCK | HAS_4BIT_BP |
HAS_SR_BP3_BIT6
--------------------------------------------------------
IMO, HAS_16BIT_SR flag is causing lock/unlock failure,
since BP bits are calculated wrong then.
I tested also for a case where I don't parse SFDP and
reverted the condition in micron_st_nor_default_init()
for 16BIT Status Register Flag. And lock/unlock fails with
same log as Non-working case.
And this mt25qu512 has 8-BIT SR as typical micron-st flash.
> Also you could enable dev_dbg to see where you get -EIO. Probably when
> reading the SR back. Also you can use debugfs to check what is set in
> the working scenario and what params are different in the non-working
> case. See drivers/mtd/spi-nor/debugfs.c
>
> Cheers,
> ta
>
>
>> -----------------------------------------------------------------
>> Case 1: BP Flags and removed size, and no_sfdp so by default expecting
>> to read SFDP
>>
>> diff --git a/drivers/mtd/spi-nor/micron-st.c
>> b/drivers/mtd/spi-nor/micron-st.c
>> index 6c8cabbead2e..4feb03ee2d13 100644
>> --- a/drivers/mtd/spi-nor/micron-st.c
>> +++ b/drivers/mtd/spi-nor/micron-st.c
>> @@ -407,9 +407,6 @@ static const struct flash_info st_nor_parts[] = {
>> .name = "mt25qu512a",
>> .flags = SPI_NOR_HAS_LOCK | SPI_NOR_HAS_TB |
>> SPI_NOR_4BIT_BP |
>> SPI_NOR_BP3_SR_BIT6,
>> - .size = SZ_64M,
>> - .no_sfdp_flags = SECT_4K | SPI_NOR_DUAL_READ |
>> SPI_NOR_QUAD_READ,
>> - .fixup_flags = SPI_NOR_4B_OPCODES,
>> .mfr_flags = USE_FSR,
>> }, {
>>
>>
>>
>>>
>>> Cheers,
>>> ta
>>>
>>>> Lock/unlock works.
>>>>
>>>> ## x86-64-smarc-evk-uwd0j0007 # uname -r
>>>> 6.6.0-rc2
>>>>
>>>> # flash_lock -i /dev/mtd0
>>>> Start: 0
>>>> Len: 0x4000000
>>>> Lock status: unlocked
>>>> Return code: 0
>>>>
>>>> # flash_lock -l /dev/mtd0
>>>> # flash_lock -i /dev/mtd0
>>>> Device: /dev/mtd0
>>>> Start: 0
>>>> Len: 0x4000000
>>>> Lock status: locked
>>>> Return code: 1
>>>>
>>>> ## x86-64-smarc-evk-uwd0j0007 # mtd_debug erase /dev/mtd0 0 1048576
>>>> [ 413.472411] spi-nor spi-PRP0001:00: Erase operation failed.
>>>> [ 413.478084] spi-nor spi-PRP0001:00: Attempted to modify a protected
>>>> sector.
>>>> MEMERASE: Input/output error
>>>>
>>>> # flash_lock -u /dev/mtd0
>>>> # flash_lock -i /dev/mtd0
>>>> Device: /dev/mtd0
>>>> Start: 0
>>>> Len: 0x4000000
>>>> Lock status: unlocked
>>>> Return code: 0
>>>>
>>>> ## x86-64-smarc-evk-uwd0j0007 # mtd_debug erase /dev/mtd0 0 1048576
>>>> Erased 1048576 bytes from address 0x00000000 in flash
>>>>
>>>>
>>>> Further I tested on stable 6.5.5 Kernel with old way of Flash Info
>>>> Format and which has forced PARSE_SFDP Flag, no issues with lock/unlock.
>>>>
>>>>
>>
Thanks,
Mamta
Powered by blists - more mailing lists