public inbox for linux-bcachefs@vger.kernel.org
 help / color / mirror / Atom feed
From: Mae Kasza <admin@badat.dev>
To: Hongbo Li <lihongbo22@huawei.com>, Mae Kasza <git@badat.dev>,
	Kent Overstreet <kent.overstreet@linux.dev>
Cc: linux-bcachefs@vger.kernel.org
Subject: Re: [PATCH] cmd_set_passphrase: initialize KDF parameters
Date: Wed, 14 Aug 2024 14:25:48 +0200	[thread overview]
Message-ID: <b31673a7-dddd-45c5-9e55-798a2ff65407@badat.dev> (raw)
In-Reply-To: <68c4ce8a-c904-4d71-9a8b-480e8d37a0e3@huawei.com>


On 8/14/24 04:10, Hongbo Li wrote:
>
>
> On 2024/8/13 21:16, Mae Kasza wrote:
>>
>> On 8/12/24 04:50, Hongbo Li wrote:
>>>
>>>
>>> On 2024/8/12 9:41, Kent Overstreet wrote:
>>>> On Sun, Aug 11, 2024 at 11:40:38PM GMT, Mae Kasza wrote:
>>>>> From: mae <git@badat.dev>
>>>>>
>>>>> The set-passphrase command failed to derive the key for disks 
>>>>> initially
>>>>> formatted with --encrypted and --no_passphrase.
>>>>>
>>>>> This happened because bch_sb_crypt_init only configures the KDF 
>>>>> params
>>>>> if a passphrase is specified.
>>>>> This commit makes the command initialize the KDF with the same 
>>>>> parameters
>>>>> as bch_sb_crypt_init if the key wasn't encrypted before.
>>>>>
>>>>> Signed-off-by: Mae Kasza <git@badat.dev>
>>>>
>>>> This looks good to me, but I'm in the middle of a 20 hour drive and 
>>>> not
>>>> feeling super confident to review crypto stuff tonight, so maybe 
>>>> someone
>>>> else can go over it too before I pull it in..
>>>>
>>>>> ---
>>>>>   c_src/cmd_key.c |  9 +++++++--
>>>>>   c_src/crypto.c  | 13 ++++++++-----
>>>>>   c_src/crypto.h  |  1 +
>>>>>   3 files changed, 16 insertions(+), 7 deletions(-)
>>>>>
>>>>> diff --git a/c_src/cmd_key.c b/c_src/cmd_key.c
>>>>> index adb0ac8d..2da83758 100644
>>>>> --- a/c_src/cmd_key.c
>>>>> +++ b/c_src/cmd_key.c
>>>>> @@ -104,7 +104,8 @@ int cmd_set_passphrase(int argc, char *argv[])
>>>>>       if (IS_ERR(c))
>>>>>           die("Error opening %s: %s", argv[1], 
>>>>> bch2_err_str(PTR_ERR(c)));
>>>>>   -    struct bch_sb_field_crypt *crypt = 
>>>>> bch2_sb_field_get(c->disk_sb.sb, crypt);
>>>>> +    struct bch_sb *sb = c->disk_sb.sb;
>>>>> +    struct bch_sb_field_crypt *crypt = bch2_sb_field_get(sb, crypt);
>>>>>       if (!crypt)
>>>>>           die("Filesystem does not have encryption enabled");
>>>>>   @@ -116,9 +117,13 @@ int cmd_set_passphrase(int argc, char *argv[])
>>>>>           die("Error getting current key");
>>>>>         char *new_passphrase = read_passphrase_twice("Enter new 
>>>>> passphrase: ");
>>>>> +    if (!bch2_key_is_encrypted(&crypt->key)) {
>>>>> +        bch_crypt_default_kdf_init(crypt);
>>> It works, but using bch_sb_crypt_init to reinitialize crypt may be 
>>> better. In bch_sb_crypt_init, it has initialized the crypt and also 
>>> cleaned new_passphrase.
>>>
>>> Thanks
>>> Hongbo
>>>
>> bch_sb_crypt_init also reinitializes the crypt->key, so I can't just 
>> reuse it here.
>>
> Yeah, it will change crypt->key. It seems the main point is not clean 
> the password.
> Do we need keep the crypt->key even the password is changed? Or does 
> the crypt->key like a random seed and need to be kept in the whole 
> lifecycle? Sorry I am a bit confused about this.
>
> Thanks,
> Hongbo

crypt->key does need to stay constant for the entire lifetime of a 
partition, since that's what's actually used to encrypt the data on the 
partition. The passphrase is just used to encrypt the crypt->key

Thanks,

Mae

>
>> I could extract it into a bch_crypt_update_passphrase function. 
>> Originally I wanted to avoid doing that to limit the scope of this 
>> patchset, but I'll give it a shot
>>
>> Thanks for the review!
>>
>> Mae
>>
>>>>> +    }
>>>>> +
>>>>>       struct bch_key passphrase_key = derive_passphrase(crypt, 
>>>>> new_passphrase);
>>>>>   -    if (bch2_chacha_encrypt_key(&passphrase_key, 
>>>>> __bch2_sb_key_nonce(c->disk_sb.sb),
>>>>> +    if (bch2_chacha_encrypt_key(&passphrase_key, 
>>>>> __bch2_sb_key_nonce(sb),
>>>>>                       &new_key, sizeof(new_key)))
>>>>>           die("error encrypting key");
>>>>>       crypt->key = new_key;
>>>>> diff --git a/c_src/crypto.c b/c_src/crypto.c
>>>>> index 32671bd8..30ad92d4 100644
>>>>> --- a/c_src/crypto.c
>>>>> +++ b/c_src/crypto.c
>>>>> @@ -180,11 +180,7 @@ void bch_sb_crypt_init(struct bch_sb *sb,
>>>>>       get_random_bytes(&crypt->key.key, sizeof(crypt->key.key));
>>>>>         if (passphrase) {
>>>>> -
>>>>> -        SET_BCH_CRYPT_KDF_TYPE(crypt, BCH_KDF_SCRYPT);
>>>>> -        SET_BCH_KDF_SCRYPT_N(crypt, ilog2(16384));
>>>>> -        SET_BCH_KDF_SCRYPT_R(crypt, ilog2(8));
>>>>> -        SET_BCH_KDF_SCRYPT_P(crypt, ilog2(16));
>>>>> +        bch_crypt_default_kdf_init(crypt);
>>>>>             struct bch_key passphrase_key = 
>>>>> derive_passphrase(crypt, passphrase);
>>>>>   @@ -199,3 +195,10 @@ void bch_sb_crypt_init(struct bch_sb *sb,
>>>>>           memzero_explicit(&passphrase_key, sizeof(passphrase_key));
>>>>>       }
>>>>>   }
>>>>> +
>>>>> +void bch_crypt_default_kdf_init(struct bch_sb_field_crypt *crypt) {
>>>>> +        SET_BCH_CRYPT_KDF_TYPE(crypt, BCH_KDF_SCRYPT);
>>>>> +        SET_BCH_KDF_SCRYPT_N(crypt, ilog2(16384));
>>>>> +        SET_BCH_KDF_SCRYPT_R(crypt, ilog2(8));
>>>>> +        SET_BCH_KDF_SCRYPT_P(crypt, ilog2(16));
>>>>> +}
>>>>> diff --git a/c_src/crypto.h b/c_src/crypto.h
>>>>> index baea6d86..846a8931 100644
>>>>> --- a/c_src/crypto.h
>>>>> +++ b/c_src/crypto.h
>>>>> @@ -18,5 +18,6 @@ void bch2_passphrase_check(struct bch_sb *, 
>>>>> const char *,
>>>>>   void bch2_add_key(struct bch_sb *, const char *, const char *, 
>>>>> const char *);
>>>>>   void bch_sb_crypt_init(struct bch_sb *sb, struct 
>>>>> bch_sb_field_crypt *,
>>>>>                  const char *);
>>>>> +void bch_crypt_default_kdf_init(struct bch_sb_field_crypt *);
>>>>>     #endif /* _CRYPTO_H */
>>>>> -- 
>>>>> 2.45.2
>>>>>
>>>>
>>>>
>>>

  reply	other threads:[~2024-08-14 12:25 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-11 21:40 Mae Kasza
2024-08-11 21:40 ` [PATCH] cmd_set_passphrase: initialize KDF parameters Mae Kasza
2024-08-12  1:41   ` Kent Overstreet
2024-08-12  2:50     ` Hongbo Li
2024-08-13 13:16       ` Mae Kasza
2024-08-13 17:37         ` [PATCH] Extract bch_crypt_update_passphrase function Mae Kasza
2024-08-14  2:21           ` Hongbo Li
2024-08-14 12:29             ` Mae Kasza
2024-08-14  2:10         ` [PATCH] cmd_set_passphrase: initialize KDF parameters Hongbo Li
2024-08-14 12:25           ` Mae Kasza [this message]
2024-09-13 21:39             ` Mae Kasza

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b31673a7-dddd-45c5-9e55-798a2ff65407@badat.dev \
    --to=admin@badat.dev \
    --cc=git@badat.dev \
    --cc=kent.overstreet@linux.dev \
    --cc=lihongbo22@huawei.com \
    --cc=linux-bcachefs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox