`hsm_secret` generation from a seed-phrase #4065

positiveblue · 2020-09-18T15:34:40Z

Add generatehsm method to hsmtool to derivate BIP32 seeds from a
mnemonic using the BIP39 standard.

The new method uses libwally for the BIP39 to BIP32 derivation. It also
fails if an hsm_secret file already exists, so we do not overwrite
someone else's wallet without noticing.

It allows the use of passphrases, the ECHO mode in the terminal is
disable for higher security.

positiveblue · 2020-09-18T15:37:29Z

Related to the issue created during the Lightning HackSprint May 2020 #3717

@ZmnSCPxj @darosior I guess that you are the right people for reviewing this

darosior

Nice, will review soon (:tm:) but strong concept ack :)

darosior

Just a quick pass as Travis does not pass yet, but wanted to give it a look!

darosior · 2020-09-22T20:50:19Z

tools/hsmtool.c

+	char word[50];
+	uint n_words;


nit: can you declare those at the top of the function?

darosior · 2020-09-22T20:50:51Z

tools/hsmtool.c

+	char word[50];
+	uint n_words;
+	for (n_words = 0; n_words < 24; n_words++) {
+		scanf("%s", word);


It does not compile (-Werror=unused-result)

darosior · 2020-09-22T20:56:07Z

tools/hsmtool.c

+	for (n_words = 0; n_words < 24; n_words++) {
+		scanf("%s", word);
+		/* We need to concat the strings using ' ' so libwally can
+		 * validate it */
+		if (n_words == 0) {
+			strcpy(mnemonic, word);
+		} else {
+			strcat(mnemonic, " ");
+			strcat(mnemonic, word);
+		}
+	}


Why not taking an entire line instead ? You could use getline as we do today for the password (but to be explicit i think we should keep the echo for the mnemonics):

lightning/lightningd/options.c

Lines 415 to 416 in a08e835

if (getline(&passwd, &passwd_size, stdin) < 0)

return "Could not read password from stdin.";

darosior · 2020-09-22T21:12:31Z

tools/hsmtool.c

+	/* Replace the last character which will always be '\n' */
+	passphrase[strlen(passphrase) - 1] = '\0';


Not always:

#include <inttypes.h> #include <stdio.h> #include <string.h> int main(int argc, char **argv) { char *test = NULL; size_t len; getline(&test, &len, stdin); printf("%s\n", test[strlen(test) - 1] == '\n' ? "true" : "false"); }

darosior@darosior:~$ gcc a.c darosior@darosior:~$ ./a.out test_pass true darosior@darosior:~$ printf passphrase_from_elsewhere |./a.out false

This is a really good catch because I did not think about that... Thanks @darosior!

Ok, after some time playing with this I think that we have a problem here... I do not see a way to know if the last character is noise or was introduced by the user. The reason is that getline returns the number of characters that where read, including the \n if present but not putting it in the read bytes. Then, if we know that we readed 2 chars, we do not know if test[1] is noise that was there because the user pushed enter \n or if the user executed printf "xc" | ./myprogram and then test[1] is the "c"

I think that the best is going to be to assume that the user is using this interactively and always replace the last character by \0 as it was the enter and was not supposed to be part of the passphrase.

If not, we have that

#include <inttypes.h> #include <stdio.h> #include <string.h> int main(int argc, char **argv) { char *test = NULL; size_t len; size_t characters = getline(&test, &len, stdin); printf("read chars: %ld\n", characters); printf("%s\n", test[strlen(test) - 1] == '\n' ? "true" : "false"); }

Then...

~/tmp ~$ ./a.out a read chars: 2 true ~/tmp ~$ printf "ab" | ./a.out read chars: 2 false

Why not doing it as we do it for --encrypted-hsm ?

lightning/lightningd/options.c

Lines 417 to 418 in cafaad7

if (passwd[strlen(passwd) - 1] == '\n')

passwd[strlen(passwd) - 1] = '\0';

EDIT: as per your example:

#include <inttypes.h> #include <stdio.h> #include <string.h> int main(int argc, char **argv) { char *test = NULL; size_t len; size_t characters = getline(&test, &len, stdin); if (test[strlen(test) - 1] == '\n') test[strlen(test) - 1] = '\0'; printf("read chars: %ld\n", characters); printf("%s\n", test[strlen(test) - 1] == '\n' ? "true" : "false"); }

darosior@darosior:~$ ./a.out a read chars: 2 false darosior@darosior:~$ printf "ab" | ./a.out read chars: 2 false

darosior · 2020-09-22T21:15:08Z

tools/hsmtool.c

@@ -33,6 +35,7 @@ static void show_usage(const char *progname)
 	       "<path/to/hsm_secret> [hsm_secret password]\n");
 	printf("	- guesstoremote <P2WPKH address> <node id> <tries> "
 	       "<path/to/hsm_secret> [hsm_secret password]\n");
+	printf("	- generatehsm <path/to/hsm_secret>\n");


Suggested change

printf(" - generatehsm <path/to/hsm_secret>\n");

printf(" - generatehsm <path/to/new//hsm_secret>\n");

To imply it's not existing yet ?

I am not used to that nomenclature but is it common for new files to be described as that?

Absolutely not, entirely baked by myself (though i'm almost sure to have read that somewhere as well) because it's clearer in my mind. :)
Was more an opinion than a request!

darosior · 2020-09-22T21:18:19Z

tools/hsmtool.c

+	if (close(fd) != 0) 
+		errx(ERROR_USAGE, "Error closing hsm_secret file");
+
+	printf("New hsm_secret file created. Use hsmtool to encrypt the BIP32 seed if needed");


Suggested change

printf("New hsm_secret file created. Use hsmtool to encrypt the BIP32 seed if needed");

printf("New hsm_secret file created. Use the `encrypt` command to encrypt the BIP32 seed if needed");

Maybe also echo the path that the hsm_secret file was created at?

New hsm_secret file created. -> New /path/to/hsm_secret created.

darosior · 2020-09-22T21:19:29Z

tools/hsmtool.c

+		/* if hsm_secred already exists we abort the process
+		 * we do not want to lose someone else's funds */
+		struct stat st;
+		if (stat(hsm_secret_path, &st) == 0)
+			errx(ERROR_USAGE, "hsm_secret file already exists");


Nice sanity check :)

I'd echo the path that already exists here also e.g.

Suggested change

/* if hsm_secred already exists we abort the process

* we do not want to lose someone else's funds */

struct stat st;

if (stat(hsm_secret_path, &st) == 0)

errx(ERROR_USAGE, "hsm_secret file already exists");

/* if hsm_secred already exists we abort the process

* we do not want to lose someone else's funds */

struct stat st;

if (stat(hsm_secret_path, &st) == 0)

errx(ERROR_USAGE, "/path/to/hsm_secret already exists. Aborting.");

positiveblue · 2020-10-05T18:26:53Z

doc/lightning-hsmtool.8

@@ -76,4 +80,4 @@ Note: the modules in the ccan/ directory have their own licenses, but
 the rest of the code is covered by the BSD-style MIT license\.
 Main web site: \fIhttps://github.com/ElementsProject/lightning\fR



yes, I did not squash it because I saw so many files changed (and not added in the gitignore that it was weird...) I will do it at the end

niftynei · 2020-10-05T17:51:14Z

tools/hsmtool.c

+	printf("Introduce the BIP39 24 word list:\n");
+	char word[50];
+	uint n_words;
+	for (n_words = 0; n_words < 24; n_words++) {


BIP39 specifies that sets of 12,15,18,21, and 24 are valid mnemonic word list lengths.

Should we support all of them? How do you want the user to define how long their mnemonic is? another input asking for the number? two empty lines in a row?

Well if you do 'readline" for an entire line which includes the complete passphrase, there's no problem with word counting? I'm not sure how robust libwally's validation method is; you might want to validate that that works as expected

niftynei · 2020-10-05T17:51:44Z

tools/hsmtool.c

@@ -368,6 +371,91 @@ static int guess_to_remote(const char *address, struct node_id *node_id,
 	return 1;
 }

+static void read_mnemonic(char *mnemonic) {
+	printf("Introduce the BIP39 24 word list:\n");


Suggested change

printf("Introduce the BIP39 24 word list:\n");

printf("Input the BIP39 word list:\n");

niftynei · 2020-10-05T17:53:24Z

tools/hsmtool.c

+		if (n_words == 0) {
+			strcpy(mnemonic, word);
+		} else {
+			strcat(mnemonic, " ");


Japanese (and Chinese?) word lists use a different word separator character. It's probably safer to have the input be a single line, instead of word tokens.

I will update it to support all the languages

niftynei · 2020-10-05T17:54:34Z

tools/hsmtool.c

+
+static void read_passphrase(char *passphrase) {
+	struct termios current_term, temp_term;
+	printf("Warning: remember that different passphrase yield different "


Suggested change

printf("Warning: remember that different passphrase yield different "

printf("Warning: remember that different passphrases yield different "

niftynei · 2020-10-05T18:00:02Z

tools/hsmtool.c

+{
+	char mnemonic[BIP39_WORDLIST_LEN];
+	read_mnemonic(mnemonic);
+	if (bip39_mnemonic_validate(NULL, mnemonic) != 0)


does this correctly validate non-English word list sets? You might try the test cases from https://github.com/bip32JP/bip32JP.github.io/blob/master/test_JP_BIP39.json

niftynei · 2020-10-05T18:01:53Z

tools/hsmtool.c

+	if (close(fd) != 0) 
+		errx(ERROR_USAGE, "Error closing hsm_secret file");
+
+	printf("New hsm_secret file created. Use hsmtool to encrypt the BIP32 seed if needed");


Maybe also echo the path that the hsm_secret file was created at?

New hsm_secret file created. -> New /path/to/hsm_secret created.

niftynei · 2020-10-05T18:02:14Z

tools/hsmtool.c

@@ -413,5 +501,20 @@ int main(int argc, char *argv[])
 		                       argv[5], argc >= 7 ? argv[6] : NULL);
 	}

+	if (streq(method, "generatehsm")) {
+		if ( argc != 3)


Suggested change

if ( argc != 3)

if (argc != 3)

niftynei · 2020-10-05T18:02:33Z

tools/hsmtool.c

+
+		char *hsm_secret_path = argv[2];
+
+		/* if hsm_secred already exists we abort the process


Suggested change

/* if hsm_secred already exists we abort the process

/* if hsm_secret already exists we abort the process

niftynei · 2020-10-05T18:03:28Z

tools/hsmtool.c

+		/* if hsm_secred already exists we abort the process
+		 * we do not want to lose someone else's funds */
+		struct stat st;
+		if (stat(hsm_secret_path, &st) == 0)
+			errx(ERROR_USAGE, "hsm_secret file already exists");


I'd echo the path that already exists here also e.g.

Suggested change

/* if hsm_secred already exists we abort the process

* we do not want to lose someone else's funds */

struct stat st;

if (stat(hsm_secret_path, &st) == 0)

errx(ERROR_USAGE, "hsm_secret file already exists");

/* if hsm_secred already exists we abort the process

* we do not want to lose someone else's funds */

struct stat st;

if (stat(hsm_secret_path, &st) == 0)

errx(ERROR_USAGE, "/path/to/hsm_secret already exists. Aborting.");

positiveblue · 2020-10-29T08:31:15Z

@niftynei I updated the PR. Now it reads the mnemonic all at once using getline() and use that directly.

If the list of words is not provided in the right (standard) format (word1 word2 word3...) it fails. Because we use libwally under the hood, we can validate and derive the seed without any problem, supporting out of the box all the mnemonic length 12, 18, 24...) and multiple languages.

I tried to compare the obtained results with the ones from other bip39 implementations and I got the same results (with and without passphrase) for english, spanish, italian, chinese... including non ASCII characters too.

However, there is one still failing. Japanese mnemonics do not pass the libwally validation. I wonder if it has something to do with the terminal byte encoding or something like that because with Chinese characters I did not have any problem... do you know if you have to do something special with JP characters? Even so, I can just show and error if Japanese is selected (or not show the option) and leave it as a TODO for the next PR.

darosior

One main request on my side: we should support passphrase == NULL instead of passphrase == '\0'.

darosior · 2020-10-29T10:28:18Z

tools/hsmtool.c

+	printf("Select your language:\n");
+	printf("  0) English (en)\n");
+	printf("  1) Spanish (es)\n");
+	printf("  2) French (fr)\n");
+	printf("  3) Italian (it)\n");
+	printf("  4) Japanese (jp)\n");
+	printf("  5) Chinese simplified (zhs)\n");
+	printf("  6) Chinese traditional (zht)\n");


Conceptually, i feel that only ever supporting English as the international language for mnemonics is safer. Don't want to block you on this though.

darosior · 2020-10-29T10:31:39Z

tools/hsmtool.c

+	printf("  6) Chinese traditional (zht)\n");
+	printf("Select [0-6]: ");
+
+	char *selection=NULL;


nit:
char *selection = NULL;

darosior · 2020-10-29T12:36:09Z

doc/lightning-hsmtool.8

@@ -51,6 +51,10 @@ and is usually no greater than the number of channels that the node has
 ever had\.
 Specify \fIpassword\fR if the \fBhsm_secret\fR is encrypted\.

+
+\fBgeneratehsm\fR \fIhsm_secret_path\fR


nit: frombip39 / secretfrombip39 / generatefrombip39 ?

The documentation says that it generates the key following the BIP39 and also the text prompted to the user during the generation... if you prefer one of those names I am happy to change it but bip39 should be the standard by now (even when lnd nodes use aezeed). In the future, if we want to support multiple "standards", we can add a flag: generatehsm --alg={aezeed, bip39,..}

darosior · 2020-10-29T12:38:46Z

tools/hsmtool.c

+		errx(ERROR_USAGE, "Could not read line from stdin.");
+	int selected = atoi(selection);
+
+	char *lang = malloc(4);


darosior · 2020-10-29T12:47:13Z

tools/hsmtool.c

+	/* Replace the last character because we assume it was the newline '\n' */
+	(*passphrase)[characters-1] = '\0';


Why do we assume it was ? Couldn't we just check it is ?

Also we should support passphrase == NULL, for the very common "no passphrase" case !

Also, characters can in theory be 0

Why do we assume it was ? Couldn't we just check it is ?

well, the user needs to introduce a value an press enter (getline()). I can add a check and fail in case something weird is going on.

Also we should support passphrase == NULL, for the very common "no passphrase" case !

The empty word and NULL have the same effect in libwally. I doubt that this never changes but I can substitute it by NULL when the user does not add any character to their passphrase.

Also, characters can in theory be 0

How do we get the line then? ctrl-d or something? In that case I think it's ok to just segfault

well, the user needs to introduce a value an press enter (getline()). I can add a check and fail in case something weird is going on.

#4065 (comment) ... I know it's harder to get there with the two consecutive getline, but that's a tool and it's going to be used in scripts and stuff.
Moreover, you are overriding the user passphrase here! This definitely needs a sanity check (maybe a belt and suspenders, but at least the belt will do for now!).
If you are really confident about this, make it an assert:

assert((*passphrase)[characters-1] == '\n'); (*passphrase)[characters-1] = '\0';

But i think it's a bad idea and can be triggered. Don't know what's wrong with just doing:

if ((*passphrase)[characters-1] == '\n') (*passphrase)[characters-1] = '\0';

The empty word and NULL have the same effect in libwally. I doubt that this never changes but I can substitute it by NULL when the user does not add any character to their passphrase.

Oh, i was not aware of that :) I just check the doc comment stating "if no passphrase then set to NULL". So yeah better to rely on documented behaviour i guess!

Check added

darosior · 2020-10-29T12:54:23Z

tools/hsmtool.c

+		default:
+			errx(ERROR_USAGE, "Invalid language selection, select one from the list [0-6].");


This cannot be reached, atoi will give you 0 if throw it garbage

You could use strtol or similar

This cannot be reached, atoi will give you 0 if throw it garbage

This is true, and I did not know it! Maybe the safest thing to do then is to take the first byte from the user's input and subtract '0' instead of doing any str to number conversion.

Hmm i'd say the safest thing here to go for strtol, which errors on --well-- error. From the manpage:

The program shown below demonstrates the use of strtol(). The first command-line argument specifies a string from which strtol() should parse a number. The second (optional) argument specifies the
base to be used for the conversion. (This argument is converted to numeric form using atoi(3), a function that performs no error checking and has a simpler interface than strtol().) Some examples of
the results produced by this program are the following:

$ ./a.out 123 strtol() returned 123 $ ./a.out ' 123' strtol() returned 123 $ ./a.out 123abc strtol() returned 123 Further characters after number: abc $ ./a.out 123abc 55 strtol: Invalid argument $ ./a.out '' No digits were found $ ./a.out 4000000000 strtol: Numerical result out of range

darosior · 2020-10-29T13:02:45Z

tools/hsmtool.c

+	char *passphrase = NULL;
+	read_passphrase(&passphrase);


If read_passphrase set passphrase, it's never freed

Thanks for this, is there any automatic tool that checks for this silly cases that I forgot? (so I can use it before future contributions and you do not have to lose your time on "simple" memory leaks)

Just went looking for missing free, but you can test with valgrind or clang with leak sanitizer enabled (configure --enable-address-sanitizer) to at least make sure the frequent paths don't leak.

Even better: write a test to trigger those codepaths :)

cdecker · 2020-10-30T19:07:51Z

tools/hsmtool.c

+		errx(ERROR_USAGE, "Could not read line from stdin.");
+
+	int selection = selected[0] - '0';
+	if (characters != 2 || selection < 0 || selection > 6)


scanf would be more appropriate here, rather than checking the length of the string and doing the char-to-int conversion.

I changed it to getline() because scanf left the \n char and sequential getline()failed. What is the best way to deal with that?

cdecker · 2020-10-30T19:08:52Z

tools/hsmtool.c

+			bip39_get_wordlist("it", words);
+			break;
+		case 4:
+			errx(ERROR_LANG_NOT_SUPPORTED, "Sorry, the language selected is not supported yet.");


It's strange to be offering an option and then declining the selection.

cdecker · 2020-10-30T19:16:23Z

tools/hsmtool.c

@@ -368,6 +372,148 @@ static int guess_to_remote(const char *address, struct node_id *node_id,
 	return 1;
 }

+static void get_words(struct words **words) {
+	printf("Select your language:\n");


There seem to be quite a number of repetitions in the code here, that might be better suited for a list of structs that contain the various pieces of information:

struct wordlist_lang { char *abbr; char *name; }; struct wordlist_lang languages[] = { {"en", "English"}, {"es", "Spanish"}, {"fr", "French"}, {"it", "Italian"}, {"jp", "Japanese"}, {"zhs", "Chinese Simplified"}, {"zht", "Chinese Traditional"}, };

Then you can use ARRAY_SIZE to determine the size of the array at compile time, iterate through them when printing the menu, and indexing with the result after selection to find the abbreviation to use when fetching the wordlist from libwally instead of having to update the printf, index checks and switch-statement below.

cdecker · 2020-10-30T19:18:40Z

tools/hsmtool.c

+	char *line = NULL;
+	size_t line_size = 0;
+
+	printf("Introduce your BIP39 word list:\n");


I'd add separated by space, so people don't attempt to write one per line, which is what I'd have done :-)

cdecker · 2020-10-30T19:23:30Z

Excellent addition @positiveblue, small nits aside, I think this is a great feature, and I'm positive we'll merge it before the next release 👍

darosior

Just two small things left, starting to look pretty good :-)

darosior · 2020-11-04T10:04:25Z

tools/hsmtool.c

+	struct termios current_term, temp_term;
+	printf("Warning: remember that different passphrase yield different "
+	       "bitcoin wallets.\n");
+	printf("Leave it empty for using the value by default (echo is "


+1, probably better to say "if left empty, no password is used".

darosior · 2020-11-04T10:29:53Z

tools/hsmtool.c

+	// If the user did not introduce any password, we want to set passphrase
+	// to NULL not to '\0' for libwally
+	if (strlen(*passphrase) == 0) {
+		free(*passphrase);
+	}


You don't set it to NULL here

free(*passphrase); *passphrase = NULL;

niftynei

Few more nits, otherwise looks good.

It occurred to me while looking into this that the "normal" BIP39 to BIP32 pathway isn't true for c-lightning. Typically you'd pass the seed generated by BIP39's derivation directly to bip32_key_from_seed. However, in hsmd.c, we add an intermediate derivation (using an hkdf), passing the BIP39 seed into this function to get yet another seed, which we then pass to bip32_key_from_seed.

This isn't a blocker for this patch. However, it will prevent "seed portability". You can't use a mnemonic here and anticipate being able to "seamlessly" import your lightning funds elsewhere using the same mnemonic.

If we want 'full compliance' with the BIP39 to BIP32 pathway, we'll need to update the seed derivation to remove this secondary step. This would be a good opportunity to more fully embrace BIP32 pathways for different key paths -- I hear LND has an (undocumented) pathway scheme for key generation and usage that might be worth porting over to c-lightning ...

niftynei · 2020-11-05T22:02:19Z

tools/hsmtool.c

+	char *endptr;
+	long val = strtol(selected, &endptr, 10);
+	if (errno == ERANGE || (errno != 0 && val == 0) || endptr == selected || val < 0 || val > ARRAY_SIZE(languages))
+        errx(ERROR_USAGE, "Invalid language selection, select one from the list [0-6].");


should be 0- ARRAY_SIZE? 😉

niftynei · 2020-11-05T22:04:55Z

tools/hsmtool.c

+	if (characters < 0)
+		errx(ERROR_USAGE, "Could not read passphrase from stdin.");
+
+	// newline is not part of the valid passphrase


style-nit: we use /* */ for comments

niftynei · 2020-11-05T22:05:16Z

tools/hsmtool.c

+	// to NULL not to '\0' for libwally
+	if (strlen(*passphrase) == 0) {
+		free(*passphrase);
+                *passphrase = NULL;


nit: spacing here is off?

niftynei · 2020-11-05T22:06:31Z

tools/hsmtool.c

+		errx(ERROR_USAGE, "Error writing secret to hsm_secret file");
+
+	if (fsync(fd) != 0)
+		errx(ERROR_USAGE, "Error fsycning hsm_secret file");


Suggested change

errx(ERROR_USAGE, "Error fsycning hsm_secret file");

errx(ERROR_USAGE, "Error fsyncing hsm_secret file");

darosior · 2020-11-06T11:37:43Z

@positiveblue feel free to ping me as soon as you force push, would like to ACK this before we cut the release :-)

This would be a good opportunity to more fully embrace BIP32 pathways for different key paths -- I hear LND has an (undocumented) pathway scheme for key generation and usage that might be worth porting over to c-lightning ...

Hehe .. #3687 :)

tools: Add `generatehsm` method to hsmtool to derivate BIP32 seeds from a mnemonic using the BIP39 standard. The new method uses libwally for the BIP39 to BIP32 derivation. It also fails if an hsm_secret file already exists, so we do not overwrite someone else's wallet without noticing. It allows the use of passphrases, the ECHO mode in the terminal is disable for higher security. It currently supports "en", "es", "fr", "it", "jp", "zhs", "zht". Changelog-Added: hsmtool: `hsm_secret` generation from a seed-phrase following BIP39.

positiveblue · 2020-11-08T22:07:56Z

@darosior I see the problems that you mention in #3687

Because many projects are already using c-lightning we can not simple ask them to close all the channels and open them again. If we really want the project to support BIP39, Azeed (lnd) and other derivations we should support the old way as a legacy feature and introduce the new third-party compatible derivations as default.

For Azeed is easy, but for BIP39 is hard taking into account that we can not use the current populate_secretstuff function form hsmd.c.

Right now people can only recover their wallet by copy the hsm_secret file. This PR could be helpful for people creating their hsm_secret in a reproducible way for the future. However, at some point the project should just use the default bip39 derivation path and add a flag for all wallets (which would use the current hkdf_sha256 step)

What do you think about it?

niftynei · 2020-11-09T23:49:10Z

ACK 7cc0674

darosior · 2020-11-10T14:10:29Z

post-merge ACK 7cc0674 . Thanks for having worked on this @positiveblue !

Right now people can only recover their wallet by copy the hsm_secret file.

This a not a good way to backup and/or recover the wallet, without the db you may (are going to?) be missing utxos. See #4171 for more discussions about that

darosior · 2021-01-02T22:48:56Z

Hmm so i'm pretty sure to have tested this with encrypt and decrypt, and even just that it yields an actual hsm_secret. But it so happens that it creates a 64-bytes hsm_secret !
I tested both master and this branch, and neither does it not seem to be libwally as it didn't change much: https://github.com/ElementsProject/libwally-core/blob/master/src/bip39.c ...

darosior reviewed Sep 18, 2020

View reviewed changes

darosior reviewed Sep 22, 2020

View reviewed changes

niftynei reviewed Oct 5, 2020

View reviewed changes

niftynei added the needs-rebase label Oct 5, 2020

niftynei added the clightning_twit Tag to nudge @niftynei to post to @clightning_twit label Oct 15, 2020

positiveblue force-pushed the plugin-bip39 branch 6 times, most recently from 91a727a to edab496 Compare October 29, 2020 01:35

positiveblue requested a review from niftynei October 29, 2020 08:18

cdecker removed the needs-rebase label Oct 29, 2020

darosior suggested changes Oct 29, 2020

View reviewed changes

positiveblue force-pushed the plugin-bip39 branch from edab496 to 039422e Compare October 29, 2020 22:34

positiveblue requested a review from darosior October 30, 2020 05:02

cdecker reviewed Oct 30, 2020

View reviewed changes

cdecker added this to the v0.9.2 milestone Oct 30, 2020

positiveblue force-pushed the plugin-bip39 branch from 039422e to 0d8da29 Compare November 1, 2020 22:17

darosior reviewed Nov 4, 2020

View reviewed changes

positiveblue force-pushed the plugin-bip39 branch from 0d8da29 to 828d3b2 Compare November 5, 2020 10:02

niftynei reviewed Nov 6, 2020

View reviewed changes

niftynei mentioned this pull request Nov 6, 2020

Remove <hsm password> entry on command line from hsmtool #4186

Closed

positiveblue force-pushed the plugin-bip39 branch from 828d3b2 to 7cc0674 Compare November 8, 2020 21:54

positiveblue requested a review from darosior November 9, 2020 01:18

niftynei merged commit fa1483a into ElementsProject:master Nov 9, 2020

niftynei mentioned this pull request Nov 9, 2020

Hsmtool: enable dumping output descriptors of onchain wallet #4171

Merged

3 tasks

niftynei removed the clightning_twit Tag to nudge @niftynei to post to @clightning_twit label Nov 17, 2020

darosior mentioned this pull request Jan 2, 2021

hsmtool: generatehsm does not generate a valid hsm_secret (anymore?) #4302

Closed

	if (getline(&passwd, &passwd_size, stdin) < 0)
	return "Could not read password from stdin.";

		/* Replace the last character which will always be '\n' */
		passphrase[strlen(passphrase) - 1] = '\0';

	if (passwd[strlen(passwd) - 1] == '\n')
	passwd[strlen(passwd) - 1] = '\0';

	printf(" - generatehsm <path/to/hsm_secret>\n");
	printf(" - generatehsm <path/to/new//hsm_secret>\n");

	printf("New hsm_secret file created. Use hsmtool to encrypt the BIP32 seed if needed");
	printf("New hsm_secret file created. Use the `encrypt` command to encrypt the BIP32 seed if needed");

		@@ -76,4 +80,4 @@ Note: the modules in the ccan/ directory have their own licenses, but
		the rest of the code is covered by the BSD-style MIT license\.
		Main web site: \fIhttps://github.com/ElementsProject/lightning\fR

	printf("Introduce the BIP39 24 word list:\n");
	printf("Input the BIP39 word list:\n");

	printf("Warning: remember that different passphrase yield different "
	printf("Warning: remember that different passphrases yield different "


		char *hsm_secret_path = argv[2];

		/* if hsm_secred already exists we abort the process

	/* if hsm_secred already exists we abort the process
	/* if hsm_secret already exists we abort the process

		/* Replace the last character because we assume it was the newline '\n' */
		(*passphrase)[characters-1] = '\0';

		default:
		errx(ERROR_USAGE, "Invalid language selection, select one from the list [0-6].");

	errx(ERROR_USAGE, "Error fsycning hsm_secret file");
	errx(ERROR_USAGE, "Error fsyncing hsm_secret file");

hsm_secret generation from a seed-phrase #4065

hsm_secret generation from a seed-phrase #4065

Conversation

positiveblue commented Sep 18, 2020

positiveblue commented Sep 18, 2020

darosior left a comment

Choose a reason for hiding this comment

darosior left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

darosior Oct 6, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

This comment was marked as outdated.

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

positiveblue commented Oct 29, 2020

darosior left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cdecker commented Oct 30, 2020

darosior left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

niftynei left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

darosior commented Nov 6, 2020

positiveblue commented Nov 8, 2020

niftynei commented Nov 9, 2020

`hsm_secret` generation from a seed-phrase #4065

`hsm_secret` generation from a seed-phrase #4065

darosior Oct 6, 2020 •

edited

Loading

niftynei left a comment •

edited

Loading

darosior commented Jan 2, 2021 •

edited

Loading