2022-04-15 13:02:42

by Md Sadre Alam

[permalink] [raw]
Subject: Re: [PATCH] mtd: rawnand: qcom: fix memory corruption that causes panic


On 4/14/2022 8:29 PM, Miquel Raynal wrote:
> WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.
>
> Hi Manivannan,
>
> [email protected] wrote on Thu, 14 Apr 2022 20:09:07 +0530:
>
>> On Thu, Apr 14, 2022 at 02:42:36PM +0200, Miquel Raynal wrote:
>>> Hi Md,
>>>
>>> [email protected] wrote on Thu, 14 Apr 2022 17:50:48 +0530:
>>>
>>>>> Hi Md,
>>>>>
>>>>> [email protected] wrote on Thu, 14 Apr 2022 11:09:33 +0530:
>>>>>
>>>>>> This patch fixes a memory corruption that occurred in the
>>>>>> nand_scan() path for Hynix nand device.
>>>>>>
>>>>>> On boot, for Hynix nand device will panic at a weird place:
>>>>>> | Unable to handle kernel NULL pointer dereference at virtual
>>>>>> address 00000070
>>>>>> | [00000070] *pgd=00000000
>>>>>> | Internal error: Oops: 5 [#1] PREEMPT SMP ARM Modules linked in:
>>>>>> | CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.17.0-01473-g13ae1769cfb0
>>>>>> #38
>>>>>> | Hardware name: Generic DT based system PC is at
>>>>>> | nandc_set_reg+0x8/0x1c LR is at qcom_nandc_command+0x20c/0x5d0
>>>>>> | pc : [<c088b74c>] lr : [<c088d9c8>] psr: 00000113
>>>>>> | sp : c14adc50 ip : c14ee208 fp : c0cc970c
>>>>>> | r10: 000000a3 r9 : 00000000 r8 : 00000040
>>>>>> | r7 : c16f6a00 r6 : 00000090 r5 : 00000004 r4 :c14ee040
>>>>>> | r3 : 00000000 r2 : 0000000b r1 : 00000000 r0 :c14ee040
>>>>>> | Flags: nzcv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none
>>>>>> | Control: 10c5387d Table: 8020406a DAC: 00000051 Register r0
>>>>>> | information: slab kmalloc-2k start c14ee000 pointer offset
>>>>>> 64 size 2048
>>>>>> | Process swapper/0 (pid: 1, stack limit = 0x(ptrval)) nandc_set_reg
>>>>>> | from qcom_nandc_command+0x20c/0x5d0 qcom_nandc_command from
>>>>>> | nand_readid_op+0x198/0x1e8 nand_readid_op from
>>>>>> | hynix_nand_has_valid_jedecid+0x30/0x78
>>>>>> | hynix_nand_has_valid_jedecid from hynix_nand_init+0xb8/0x454
>>>>>> | hynix_nand_init from nand_scan_with_ids+0xa30/0x14a8
>>>>>> | nand_scan_with_ids from qcom_nandc_probe+0x648/0x7b0
>>>>>> | qcom_nandc_probe from platform_probe+0x58/0xac
>>>>>>
>>>>>> The problem is that the nand_scan()'s qcom_nand_attach_chip callback
>>>>>> is updating the nandc->max_cwperpage from 1 to 4.This causes the
>>>>>> sg_init_table of clear_bam_transaction() in the driver's
>>>>>> qcom_nandc_command() to memset much more than what was initially
>>>>>> allocated by alloc_bam_transaction().
>>>>> Thanks for investigating!
>>>>>
>>>>>> This patch will update nandc->max_cwperpage 1 to 4 after nand_scan()
>>>>>> returns, and remove updating nandc->max_cwperpage from
>>>>>> qcom_nand_attach_chip call back.
>>>>> The fix does not look right, as far as I understand, this should be properly handled during the attach phase. That is where we have all information about the chip and do the configuration for this chip.
>>>>>
>>>>> If you update max_cwperpage there you should probably update other internal variables that depend on it as well.
>>>> Currently we are updating max_cwperpage in qcom_nand_attach_chip(), but we are seeing issue for Hynix nand device since nand_scan_tail() is getting called after nand_attach() and in nand_attach() we are updating max_cwperpage to 4 or 8 based on page size.
>>>>
>>>> From nand_scan_tail() there is a call for nand_manufacturer_init() , specific to Hynix nand read_id is getting called that's why we are seeing this issue only for Hynix nand device. Read id sequence as below
>>>>
>>>> hynix_nand_has_valid_jedecid()
>>>>
>>>> |
>>>>
>>>> nand_readid_op()
>>>>
>>>> |
>>>>
>>>> qcom_nandc_command()
>>>>
>>>> |
>>>>
>>>> pre_command()
>>>>
>>>> |
>>>>
>>>> clear_bam_transaction() --> In this call we are doing sg_init_table() which is calling memset() based on max_cwperpage.Since initially we have allocated bam transaction as per max_cwperpage =1 and , since nand_chip_attach() updated max_cwperpage, now we are doing memset as per max_cwperpage = 4 or 8.
>>>>
>>>>
>>>> So anyway we have to updated max_cwperpage after nand_scan() call only. Since there is no other dependency on max_cwperpage in nand_attach_chip() and we are using this in bam_alloc() and bam_clear().
>>> Why don't you update the sg table after increasing max_cwperpage?
>>>
>> Or we could move the bam reallocation inside qcom_nand_attach_chip() as below?
> Much better approach, yes.
 Updated in V2 patch , as Manivannan suggested.
>
>> diff --git a/drivers/mtd/nand/raw/qcom_nandc.c b/drivers/mtd/nand/raw/qcom_nandc.c
>> index 7c6efa3b6255..58c16054630f 100644
>> --- a/drivers/mtd/nand/raw/qcom_nandc.c
>> +++ b/drivers/mtd/nand/raw/qcom_nandc.c
>> @@ -2653,9 +2653,23 @@ static int qcom_nand_attach_chip(struct nand_chip *chip)
>>
>> mtd_set_ooblayout(mtd, &qcom_nand_ooblayout_ops);
>>
>> + /* Free the initially allocated BAM transaction for reading the ONFI params */
>> + if (nandc->props->is_bam)
>> + free_bam_transaction(nandc);
>> +
>> nandc->max_cwperpage = max_t(unsigned int, nandc->max_cwperpage,
>> cwperpage);
>>
>> + /* Now allocate the BAM transaction based on updated max_cwperpage */
>> + if (nandc->props->is_bam) {
>> + nandc->bam_txn = alloc_bam_transaction(nandc);
>> + if (!nandc->bam_txn) {
>> + dev_err(nandc->dev,
>> + "failed to allocate bam transaction\n");
>> + return -ENOMEM;
>> + }
>> + }
>> +
>> /*
>> * DATA_UD_BYTES varies based on whether the read/write command protects
>> * spare data with ECC too. We protect spare data by default, so we set
>> @@ -2956,17 +2970,6 @@ static int qcom_nand_host_init_and_register(struct qcom_nand_controller *nandc,
>> if (ret)
>> return ret;
>>
>> - if (nandc->props->is_bam) {
>> - free_bam_transaction(nandc);
>> - nandc->bam_txn = alloc_bam_transaction(nandc);
>> - if (!nandc->bam_txn) {
>> - dev_err(nandc->dev,
>> - "failed to allocate bam transaction\n");
>> - nand_cleanup(chip);
>> - return -ENOMEM;
>> - }
>> - }
>> -
>> ret = mtd_device_parse_register(mtd, probes, NULL, NULL, 0);
>> if (ret)
>> nand_cleanup(chip);
>>
>> Thanks,
>> Mani
>
> Thanks,
> Miquèl