core-lightning/lightningd/pay.c

908 lines
26 KiB
C
Raw Normal View History

#include "pay.h"
#include <bitcoin/preimage.h>
#include <ccan/str/hex/hex.h>
#include <ccan/structeq/structeq.h>
#include <ccan/tal/str/str.h>
#include <common/bolt11.h>
#include <gossipd/gen_gossip_wire.h>
#include <lightningd/chaintopology.h>
#include <lightningd/jsonrpc.h>
#include <lightningd/jsonrpc_errors.h>
#include <lightningd/lightningd.h>
#include <lightningd/log.h>
#include <lightningd/options.h>
#include <lightningd/peer_control.h>
#include <lightningd/peer_htlcs.h>
#include <lightningd/subd.h>
#include <sodium/randombytes.h>
/*-----------------------------------------------------------------------------
Internal sendpay interface
-----------------------------------------------------------------------------*/
/* sendpay command */
struct sendpay_command {
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
struct list_node list;
struct sha256 payment_hash;
void (*cb)(const struct sendpay_result *, void*);
void *cbarg;
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
};
static void destroy_sendpay_command(struct sendpay_command *pc)
{
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
list_del(&pc->list);
}
/* Owned by cxt; if cxt is deleted, then cb will
* no longer be called. */
static struct sendpay_command *
new_sendpay_command(const tal_t *cxt,
const struct sha256 *payment_hash,
struct lightningd *ld,
void (*cb)(const struct sendpay_result *, void*),
void *cbarg)
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
{
struct sendpay_command *pc = tal(cxt, struct sendpay_command);
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
pc->payment_hash = *payment_hash;
pc->cb = cb;
pc->cbarg = cbarg;
list_add(&ld->sendpay_commands, &pc->list);
tal_add_destructor(pc, destroy_sendpay_command);
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
return pc;
}
/* Caller responsible for freeing ctx. */
static void sendpay_resolve(const tal_t *ctx,
struct lightningd *ld,
const struct sha256 *payment_hash,
const struct sendpay_result *result)
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
{
struct sendpay_command *pc;
struct sendpay_command *next;
list_for_each_safe(&ld->sendpay_commands, pc, next, list) {
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
if (!structeq(payment_hash, &pc->payment_hash))
continue;
/* Delete later (in our own caller) if callback did
* not delete. */
tal_steal(ctx, pc);
pc->cb(result, pc->cbarg);
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
}
}
static void sendpay_success(struct lightningd *ld,
const struct sha256 *payment_hash,
const struct preimage *payment_preimage)
{
const tal_t *tmpctx = tal_tmpctx(ld);
struct sendpay_result *result;
result = tal(tmpctx, struct sendpay_result);
result->succeeded = true;
result->preimage = *payment_preimage;
sendpay_resolve(tmpctx, ld, payment_hash, result);
tal_free(tmpctx);
}
static void sendpay_route_failure(struct lightningd *ld,
const struct sha256 *payment_hash,
bool retry_plausible,
struct routing_failure *fail,
const u8 *onionreply,
const char *details)
{
const tal_t *tmpctx = tal_tmpctx(ld);
struct sendpay_result *result;
result = tal(tmpctx, struct sendpay_result);
result->succeeded = false;
result->errorcode =
(!fail) ? PAY_UNPARSEABLE_ONION :
(!retry_plausible) ? PAY_DESTINATION_PERM_FAIL :
/*otherwise*/ PAY_TRY_OTHER_ROUTE ;
result->onionreply = onionreply;
result->routing_failure = fail;
result->details = details;
sendpay_resolve(tmpctx, ld, payment_hash, result);
tal_free(tmpctx);
}
/* Immediately fail during send_payment call. */
static void sendpay_fail_now(void (*cb)(const struct sendpay_result *, void*),
void *cbarg,
int errorcode,
char const *details)
{
const tal_t *tmpctx = tal_tmpctx(NULL);
struct sendpay_result *result;
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
result = tal(tmpctx, struct sendpay_result);
result->succeeded = false;
result->errorcode = errorcode;
result->details = details;
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
cb(result, cbarg);
tal_free(tmpctx);
}
/* Immediately fail during send_payment call. */
static void
sendpay_succeed_now(void (*cb)(const struct sendpay_result*, void*),
void *cbarg,
const struct preimage *payment_preimage)
{
const tal_t *tmpctx = tal_tmpctx(NULL);
struct sendpay_result *result;
result = tal(tmpctx, struct sendpay_result);
result->succeeded = true;
result->preimage = *payment_preimage;
cb(result, cbarg);
tal_free(tmpctx);
}
void payment_succeeded(struct lightningd *ld, struct htlc_out *hout,
const struct preimage *rval)
{
wallet_payment_set_status(ld->wallet, &hout->payment_hash,
PAYMENT_COMPLETE, rval);
sendpay_success(ld, &hout->payment_hash, rval);
}
/* Return NULL if the wrapped onion error message has no
* channel_update field, or return the embedded
* channel_update message otherwise. */
static u8 *channel_update_from_onion_error(const tal_t *ctx,
const u8 *onion_message)
{
u8 *channel_update = NULL;
u64 unused64;
u32 unused32;
/* Identify failcodes that have some channel_update.
*
* TODO > BOLT 1.0: Add new failcodes when updating to a
* new BOLT version. */
if (!fromwire_temporary_channel_failure(ctx,
onion_message,
&channel_update) &&
!fromwire_amount_below_minimum(ctx,
onion_message, &unused64,
&channel_update) &&
!fromwire_fee_insufficient(ctx,
onion_message, &unused64,
&channel_update) &&
!fromwire_incorrect_cltv_expiry(ctx,
onion_message, &unused32,
&channel_update) &&
!fromwire_expiry_too_soon(ctx,
onion_message,
&channel_update))
/* No channel update. */
channel_update = NULL;
return channel_update;
}
/* Return a struct routing_failure for an immediate failure
* (returned directly from send_htlc_out). The returned
* failure is allocated from the given context. */
static struct routing_failure*
immediate_routing_failure(const tal_t *ctx,
const struct lightningd *ld,
enum onion_type failcode,
const struct short_channel_id *channel0)
{
struct routing_failure *routing_failure;
assert(failcode);
routing_failure = tal(ctx, struct routing_failure);
routing_failure->erring_index = 0;
routing_failure->failcode = failcode;
routing_failure->erring_node = ld->id;
routing_failure->erring_channel = *channel0;
routing_failure->channel_update = NULL;
return routing_failure;
}
/* Return a struct routing_failure for a local failure allocated
* from the given context. */
static struct routing_failure*
local_routing_failure(const tal_t *ctx,
const struct lightningd *ld,
const struct htlc_out *hout,
const struct wallet_payment *payment)
{
struct routing_failure *routing_failure;
assert(hout->failcode);
routing_failure = tal(ctx, struct routing_failure);
routing_failure->erring_index = 0;
routing_failure->failcode = hout->failcode;
routing_failure->erring_node = ld->id;
routing_failure->erring_channel = payment->route_channels[0];
routing_failure->channel_update = NULL;
return routing_failure;
}
/* Return false if permanent failure at the destination, true if
* retrying is plausible. Fill *routing_failure with NULL if
* we cannot report the remote failure, or with the routing
* failure to report (allocated from ctx) otherwise. */
static struct routing_failure*
remote_routing_failure(const tal_t *ctx,
bool *p_retry_plausible,
bool *p_report_to_gossipd,
const struct wallet_payment *payment,
const struct onionreply *failure)
{
enum onion_type failcode = fromwire_peektype(failure->msg);
u8 *channel_update;
struct routing_failure *routing_failure;
const struct pubkey *route_nodes;
const struct pubkey *erring_node;
const struct short_channel_id *route_channels;
const struct short_channel_id *erring_channel;
static const struct short_channel_id dummy_channel = { 0 };
int origin_index;
bool retry_plausible;
bool report_to_gossipd;
routing_failure = tal(ctx, struct routing_failure);
route_nodes = payment->route_nodes;
route_channels = payment->route_channels;
origin_index = failure->origin_index;
channel_update
= channel_update_from_onion_error(routing_failure,
failure->msg);
retry_plausible = true;
report_to_gossipd = true;
assert(origin_index < tal_count(route_nodes));
/* Check if at destination. */
if (origin_index == tal_count(route_nodes) - 1) {
erring_channel = &dummy_channel;
/* BOLT #4:
*
* - if the _final node_ is returning the error:
* - if the PERM bit is set:
* - SHOULD fail the payment.
* */
if (failcode & PERM)
retry_plausible = false;
else
retry_plausible = true;
/* Only send message to gossipd if NODE error;
* there is no "next" channel to report as
* failing if this is the last node. */
if (failcode & NODE)
report_to_gossipd = true;
else
report_to_gossipd = false;
} else
/* Report the *next* channel as failing. */
erring_channel = &route_channels[origin_index + 1];
erring_node = &route_nodes[origin_index];
routing_failure->erring_index = (unsigned int) (origin_index + 1);
routing_failure->failcode = failcode;
routing_failure->erring_node = *erring_node;
routing_failure->erring_channel = *erring_channel;
routing_failure->channel_update = channel_update;
*p_retry_plausible = retry_plausible;
*p_report_to_gossipd = report_to_gossipd;
return routing_failure;
}
static void random_mark_channel_unroutable(struct log *log,
struct subd *gossip,
struct short_channel_id *route_channels)
{
const tal_t *tmpctx = tal_tmpctx(gossip);
size_t num_channels = tal_count(route_channels);
size_t i;
const struct short_channel_id *channel;
u8 *msg;
assert(num_channels != 0);
/* Select one channel by random. */
randombytes_buf(&i, sizeof(i));
i = i % num_channels;
channel = &route_channels[i];
log_debug(log,
"Disable randomly %dth channel (%s) along route "
"(guessing due to bad reply)",
(int) i,
type_to_string(tmpctx, struct short_channel_id,
channel));
msg = towire_gossip_mark_channel_unroutable(tmpctx, channel);
subd_send_msg(gossip, msg);
tal_free(tmpctx);
}
static void report_routing_failure(struct log *log,
struct subd *gossip,
struct routing_failure *fail)
{
const tal_t *tmpctx = tal_tmpctx(gossip);
u8 *gossip_msg;
assert(fail);
log_debug(log,
"Reporting route failure to gossipd: 0x%04x (%s) "
"node %s channel %s update %s",
fail->failcode, onion_type_name(fail->failcode),
type_to_string(tmpctx, struct pubkey,
&fail->erring_node),
type_to_string(tmpctx, struct short_channel_id,
&fail->erring_channel),
tal_hex(tmpctx, fail->channel_update));
gossip_msg = towire_gossip_routing_failure(tmpctx,
&fail->erring_node,
&fail->erring_channel,
(u16) fail->failcode,
fail->channel_update);
subd_send_msg(gossip, gossip_msg);
tal_free(tmpctx);
}
void payment_failed(struct lightningd *ld, const struct htlc_out *hout,
const char *localfail)
{
struct onionreply *reply;
enum onion_type failcode;
struct secret *path_secrets;
struct wallet_payment *payment;
const tal_t *tmpctx = tal_tmpctx(ld);
struct routing_failure* fail = NULL;
const char *failmsg;
bool retry_plausible;
bool report_to_gossipd;
payment = wallet_payment_by_hash(tmpctx, ld->wallet,
&hout->payment_hash);
#ifdef COMPAT_V052
/* FIXME: Prior to 299b280f7, we didn't put route_nodes and
* route_channels in db. If this happens, it's an old payment,
* so we can simply mark it failed in db and return. */
if (!payment->route_channels) {
log_unusual(hout->key.channel->log,
"No route_channels for htlc %s:"
" was this an old database?",
type_to_string(ltmp, struct sha256,
&hout->payment_hash));
wallet_payment_set_status(ld->wallet, &hout->payment_hash,
PAYMENT_FAILED, NULL);
tal_free(tmpctx);
return;
}
#endif
/* This gives more details than a generic failure message */
if (localfail) {
fail = local_routing_failure(tmpctx, ld, hout, payment);
failcode = fail->failcode;
failmsg = localfail;
retry_plausible = true;
report_to_gossipd = true;
} else {
/* Must be remote fail. */
assert(!hout->failcode);
failmsg = "reply from remote";
/* Try to parse reply. */
path_secrets = payment->path_secrets;
reply = unwrap_onionreply(tmpctx, path_secrets,
tal_count(path_secrets),
hout->failuremsg);
if (!reply) {
log_info(hout->key.channel->log,
"htlc %"PRIu64" failed with bad reply (%s)",
hout->key.id,
tal_hex(ltmp, hout->failuremsg));
/* Cannot report failure. */
fail = NULL;
failcode = WIRE_PERMANENT_NODE_FAILURE;
/* Select a channel to mark unroutable by random */
random_mark_channel_unroutable(hout->key.channel->log,
ld->gossip,
payment->route_channels);
/* Can now retry; we selected a channel to mark
* unroutable by random */
retry_plausible = true;
/* Already reported something to gossipd, do not
* report anything else */
report_to_gossipd = false;
} else {
failcode = fromwire_peektype(reply->msg);
log_info(hout->key.channel->log,
"htlc %"PRIu64" "
"failed from %ith node "
"with code 0x%04x (%s)",
hout->key.id,
reply->origin_index,
failcode, onion_type_name(failcode));
fail = remote_routing_failure(tmpctx,
&retry_plausible,
&report_to_gossipd,
payment, reply);
}
}
/* This may invalidated the payment structure returned, so
* access to payment object should not be done after the
* below call. */
wallet_payment_set_status(ld->wallet, &hout->payment_hash,
PAYMENT_FAILED, NULL);
/* Report to gossipd if we decided we should. */
if (report_to_gossipd)
report_routing_failure(ld->log, ld->gossip, fail);
/* Report to client. */
sendpay_route_failure(ld, &hout->payment_hash,
retry_plausible, fail, hout->failuremsg,
failmsg);
tal_free(tmpctx);
}
/* Returns false if we called callback directly, true if
* callback is scheduled for later.
*
* This call expects that if it calls the callback, then
* the given context should have been freed. */
2018-02-15 04:32:03 +01:00
bool send_payment(const tal_t *ctx,
struct lightningd* ld,
const struct sha256 *rhash,
const struct route_hop *route,
void (*cb)(const struct sendpay_result*, void*),
void *cbarg)
{
const u8 *onion;
u8 sessionkey[32];
unsigned int base_expiry;
struct onionpacket *packet;
struct secret *path_secrets;
enum onion_type failcode;
const tal_t *tmpctx = tal_tmpctx(ctx);
size_t i, n_hops = tal_count(route);
struct hop_data *hop_data = tal_arr(tmpctx, struct hop_data, n_hops);
struct pubkey *ids = tal_arr(tmpctx, struct pubkey, n_hops);
struct wallet_payment *payment = NULL;
struct htlc_out *hout;
struct short_channel_id *channels;
struct routing_failure *fail;
struct channel *channel;
/* Expiry for HTLCs is absolute. And add one to give some margin. */
base_expiry = get_block_height(ld->topology) + 1;
/* Extract IDs for each hop: create_onionpacket wants array. */
for (i = 0; i < n_hops; i++)
ids[i] = route[i].nodeid;
/* Copy hop_data[n] from route[n+1] (ie. where it goes next) */
for (i = 0; i < n_hops - 1; i++) {
hop_data[i].realm = 0;
hop_data[i].channel_id = route[i+1].channel_id;
hop_data[i].amt_forward = route[i+1].amount;
hop_data[i].outgoing_cltv = base_expiry + route[i+1].delay;
}
/* And finally set the final hop to the special values in
* BOLT04 */
hop_data[i].realm = 0;
hop_data[i].outgoing_cltv = base_expiry + route[i].delay;
memset(&hop_data[i].channel_id, 0, sizeof(struct short_channel_id));
hop_data[i].amt_forward = route[i].amount;
/* Now, do we already have a payment? */
payment = wallet_payment_by_hash(tmpctx, ld->wallet, rhash);
if (payment) {
/* FIXME: We should really do something smarter here! */
log_debug(ld->log, "send_payment: found previous");
if (payment->status == PAYMENT_PENDING) {
log_add(ld->log, "Payment is still in progress");
sendpay_fail_now(cb, cbarg, PAY_IN_PROGRESS,
"Payment is still in progress");
return false;
}
if (payment->status == PAYMENT_COMPLETE) {
log_add(ld->log, "... succeeded");
/* Must match successful payment parameters. */
if (payment->msatoshi != hop_data[n_hops-1].amt_forward) {
char *msg = tal_fmt(tmpctx,
"Already succeeded "
"with amount %"PRIu64,
payment->msatoshi);
sendpay_fail_now(cb, cbarg,
PAY_RHASH_ALREADY_USED,
msg);
return false;
}
if (!structeq(&payment->destination, &ids[n_hops-1])) {
char *msg = tal_fmt(tmpctx,
"Already succeeded to %s",
type_to_string(tmpctx,
struct pubkey,
&payment->destination));
sendpay_fail_now(cb, cbarg,
PAY_RHASH_ALREADY_USED,
msg);
return false;
}
sendpay_succeed_now(cb, cbarg,
payment->payment_preimage);
return false;
}
wallet_payment_delete(ld->wallet, rhash);
log_add(ld->log, "... retrying");
}
/* At this point we know there is no duplicate payment.
* Register it to the lightningd. Use the caller
* context, not our temporary context. */
new_sendpay_command(ctx, rhash, ld, cb, cbarg);
channel = active_channel_by_id(ld, &ids[0], NULL);
if (!channel) {
/* Report routing failure to gossipd */
fail = immediate_routing_failure(tmpctx, ld,
WIRE_UNKNOWN_NEXT_PEER,
&route[0].channel_id);
report_routing_failure(ld->log, ld->gossip, fail);
/* Report routing failure to user */
sendpay_route_failure(ld, rhash, true, fail, NULL,
"No connection to first "
"peer found");
return false;
}
randombytes_buf(&sessionkey, sizeof(sessionkey));
/* Onion will carry us from first peer onwards. */
packet = create_onionpacket(tmpctx, ids, hop_data, sessionkey, rhash->u.u8,
sizeof(struct sha256), &path_secrets);
onion = serialize_onionpacket(tmpctx, packet);
log_info(ld->log, "Sending %u over %zu hops to deliver %u",
route[0].amount, n_hops, route[n_hops-1].amount);
failcode = send_htlc_out(channel, route[0].amount,
base_expiry + route[0].delay,
pay: remove cmd pointer from htlc_out. Maintaining it was always fraught, since the command could go away if the JSON RPC died. Most recently, it was broken again on shutdown (see below). In future we may allow pay commands to block on previous payments, so it won't even be a 1:1 mapping. Generalize it: keep commands in a simple list and do a lookup when a payment fails/succeeds. Valgrind error file: valgrind-errors.5732 ==5732== Invalid read of size 8 ==5732== at 0x4149FD: remove_cmd_from_hout (pay.c:292) ==5732== by 0x468BAB: notify (tal.c:237) ==5732== by 0x469077: del_tree (tal.c:400) ==5732== by 0x4690C7: del_tree (tal.c:410) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x40F1EA: main (lightningd.c:362) ==5732== Address 0x69df148 is 1,512 bytes inside a block of size 1,544 free'd ==5732== at 0x4C2EDEB: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x469150: del_tree (tal.c:421) ==5732== by 0x46948A: tal_free (tal.c:509) ==5732== by 0x4198F2: free_htlcs (peer_control.c:1281) ==5732== by 0x40EBA9: shutdown_subdaemons (lightningd.c:209) ==5732== by 0x40F1DE: main (lightningd.c:360) ==5732== Block was alloc'd at ==5732== at 0x4C2DB8F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==5732== by 0x468C30: allocate (tal.c:250) ==5732== by 0x4691F7: tal_alloc_ (tal.c:448) ==5732== by 0x40A279: new_htlc_out (htlc_end.c:143) ==5732== by 0x41FD64: send_htlc_out (peer_htlcs.c:397) ==5732== by 0x41511C: send_payment (pay.c:388) ==5732== by 0x41589E: json_sendpay (pay.c:513) ==5732== by 0x40D9B1: parse_request (jsonrpc.c:600) ==5732== by 0x40DCAC: read_json (jsonrpc.c:667) ==5732== by 0x45C706: next_plan (io.c:59) ==5732== by 0x45D1DD: do_plan (io.c:387) ==5732== by 0x45D21B: io_ready (io.c:397) Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
2018-02-02 01:55:06 +01:00
rhash, onion, NULL, &hout);
if (failcode) {
/* Report routing failure to gossipd */
fail = immediate_routing_failure(tmpctx, ld,
failcode,
&route[0].channel_id);
report_routing_failure(ld->log, ld->gossip, fail);
/* Report routing failure to user */
sendpay_route_failure(ld, rhash, true, fail, NULL,
"First peer not ready");
return false;
}
/* Copy channels used along the route. */
channels = tal_arr(tmpctx, struct short_channel_id, n_hops);
for (i = 0; i < n_hops; ++i)
channels[i] = route[i].channel_id;
/* If hout fails, payment should be freed too. */
payment = tal(hout, struct wallet_payment);
payment->id = 0;
payment->payment_hash = *rhash;
payment->destination = ids[n_hops - 1];
payment->status = PAYMENT_PENDING;
payment->msatoshi = route[n_hops-1].amount;
payment->timestamp = time_now().ts.tv_sec;
payment->payment_preimage = NULL;
payment->path_secrets = tal_steal(payment, path_secrets);
payment->route_nodes = tal_steal(payment, ids);
payment->route_channels = tal_steal(payment, channels);
/* We write this into db when HTLC is actually sent. */
wallet_payment_setup(ld->wallet, payment);
tal_free(tmpctx);
return true;
}
/*-----------------------------------------------------------------------------
JSON-RPC sendpay interface
-----------------------------------------------------------------------------*/
static void
json_sendpay_success(struct command *cmd,
const struct preimage *payment_preimage)
{
struct json_result *response;
response = new_json_result(cmd);
json_object_start(response, NULL);
json_add_hex(response, "payment_preimage",
payment_preimage, sizeof(*payment_preimage));
json_object_end(response);
command_success(cmd, response);
}
static void json_sendpay_on_resolve(const struct sendpay_result *r,
void *vcmd)
{
struct command *cmd = (struct command*) vcmd;
struct json_result *data = NULL;
const char *msg = NULL;
struct routing_failure *fail;
if (r->succeeded)
json_sendpay_success(cmd, &r->preimage);
else {
switch (r->errorcode) {
case PAY_IN_PROGRESS:
case PAY_RHASH_ALREADY_USED:
data = NULL;
msg = r->details;
break;
case PAY_UNPARSEABLE_ONION:
data = new_json_result(cmd);
json_object_start(data, NULL);
json_add_hex(data, "onionreply",
r->onionreply, tal_len(r->onionreply));
json_object_end(data);
assert(r->details != NULL);
msg = tal_fmt(cmd,
"failed: WIRE_PERMANENT_NODE_FAILURE "
"(%s)",
r->details);
break;
case PAY_DESTINATION_PERM_FAIL:
case PAY_TRY_OTHER_ROUTE:
fail = r->routing_failure;
data = new_json_result(cmd);
json_object_start(data, NULL);
json_add_num(data, "erring_index",
fail->erring_index);
json_add_num(data, "failcode",
(unsigned) fail->failcode);
json_add_hex(data, "erring_node",
&fail->erring_node,
sizeof(fail->erring_node));
json_add_short_channel_id(data, "erring_channel",
&fail->erring_channel);
if (fail->channel_update)
json_add_hex(data, "channel_update",
fail->channel_update,
tal_len(fail->channel_update));
json_object_end(data);
assert(r->details != NULL);
msg = tal_fmt(cmd,
"failed: %s (%s)",
onion_type_name(fail->failcode),
r->details);
break;
}
assert(msg);
command_fail_detailed(cmd, r->errorcode, data, "%s", msg);
}
}
static void json_sendpay(struct command *cmd,
const char *buffer, const jsmntok_t *params)
{
jsmntok_t *routetok, *rhashtok;
const jsmntok_t *t, *end;
size_t n_hops;
struct sha256 rhash;
struct route_hop *route;
if (!json_get_params(cmd, buffer, params,
"route", &routetok,
"payment_hash", &rhashtok,
NULL)) {
return;
}
if (!hex_decode(buffer + rhashtok->start,
rhashtok->end - rhashtok->start,
&rhash, sizeof(rhash))) {
command_fail(cmd, "'%.*s' is not a valid sha256 hash",
2018-02-21 11:29:57 +01:00
rhashtok->end - rhashtok->start,
buffer + rhashtok->start);
return;
}
if (routetok->type != JSMN_ARRAY) {
command_fail(cmd, "'%.*s' is not an array",
2018-02-21 11:29:57 +01:00
routetok->end - routetok->start,
buffer + routetok->start);
return;
}
end = json_next(routetok);
n_hops = 0;
route = tal_arr(cmd, struct route_hop, n_hops);
for (t = routetok + 1; t < end; t = json_next(t)) {
const jsmntok_t *amttok, *idtok, *delaytok, *chantok;
if (t->type != JSMN_OBJECT) {
command_fail(cmd, "Route %zu '%.*s' is not an object",
n_hops,
2018-02-21 11:29:57 +01:00
t->end - t->start,
buffer + t->start);
return;
}
amttok = json_get_member(buffer, t, "msatoshi");
idtok = json_get_member(buffer, t, "id");
delaytok = json_get_member(buffer, t, "delay");
chantok = json_get_member(buffer, t, "channel");
if (!amttok || !idtok || !delaytok || !chantok) {
command_fail(cmd, "Route %zu needs msatoshi/id/channel/delay",
n_hops);
return;
}
tal_resize(&route, n_hops + 1);
/* What that hop will forward */
if (!json_tok_number(buffer, amttok, &route[n_hops].amount)) {
command_fail(cmd, "Route %zu invalid msatoshi",
n_hops);
return;
}
if (!json_tok_short_channel_id(buffer, chantok,
&route[n_hops].channel_id)) {
command_fail(cmd, "Route %zu invalid channel_id", n_hops);
return;
}
if (!json_tok_pubkey(buffer, idtok, &route[n_hops].nodeid)) {
command_fail(cmd, "Route %zu invalid id", n_hops);
return;
}
if (!json_tok_number(buffer, delaytok, &route[n_hops].delay)) {
command_fail(cmd, "Route %zu invalid delay", n_hops);
return;
}
n_hops++;
}
if (n_hops == 0) {
command_fail(cmd, "Empty route");
return;
}
if (send_payment(cmd, cmd->ld, &rhash, route,
&json_sendpay_on_resolve, cmd))
command_still_pending(cmd);
}
static const struct json_command sendpay_command = {
"sendpay",
json_sendpay,
"Send along {route} in return for preimage of {payment_hash}"
};
AUTODATA(json_command, &sendpay_command);
static void json_listpayments(struct command *cmd, const char *buffer,
const jsmntok_t *params)
{
const struct wallet_payment **payments;
struct json_result *response = new_json_result(cmd);
jsmntok_t *bolt11tok, *rhashtok;
struct sha256 *rhash = NULL;
if (!json_get_params(cmd, buffer, params,
"?bolt11", &bolt11tok,
"?payment_hash", &rhashtok,
NULL)) {
return;
}
if (bolt11tok) {
struct bolt11 *b11;
char *b11str, *fail;
if (rhashtok) {
command_fail(cmd, "Can only specify one of"
" {bolt11} or {payment_hash}");
return;
}
b11str = tal_strndup(cmd, buffer + bolt11tok->start,
bolt11tok->end - bolt11tok->start);
b11 = bolt11_decode(cmd, b11str, NULL, &fail);
if (!b11) {
command_fail(cmd, "Invalid bolt11: %s", fail);
return;
}
rhash = &b11->payment_hash;
} else if (rhashtok) {
rhash = tal(cmd, struct sha256);
if (!hex_decode(buffer + rhashtok->start,
rhashtok->end - rhashtok->start,
rhash, sizeof(*rhash))) {
command_fail(cmd, "'%.*s' is not a valid sha256 hash",
2018-02-21 11:29:57 +01:00
rhashtok->end - rhashtok->start,
buffer + rhashtok->start);
return;
}
}
payments = wallet_payment_list(cmd, cmd->ld->wallet, rhash);
json_object_start(response, NULL);
json_array_start(response, "payments");
for (int i=0; i<tal_count(payments); i++) {
const struct wallet_payment *t = payments[i];
json_object_start(response, NULL);
json_add_u64(response, "id", t->id);
json_add_hex(response, "payment_hash", &t->payment_hash, sizeof(t->payment_hash));
json_add_pubkey(response, "destination", &t->destination);
json_add_u64(response, "msatoshi", t->msatoshi);
if (deprecated_apis)
json_add_u64(response, "timestamp", t->timestamp);
json_add_u64(response, "created_at", t->timestamp);
switch (t->status) {
case PAYMENT_PENDING:
json_add_string(response, "status", "pending");
break;
case PAYMENT_COMPLETE:
json_add_string(response, "status", "complete");
break;
case PAYMENT_FAILED:
json_add_string(response, "status", "failed");
break;
}
if (t->payment_preimage)
json_add_hex(response, "payment_preimage",
t->payment_preimage,
sizeof(*t->payment_preimage));
json_object_end(response);
}
json_array_end(response);
json_object_end(response);
command_success(cmd, response);
}
static const struct json_command listpayments_command = {
"listpayments",
json_listpayments,
"Show outgoing payments"
};
AUTODATA(json_command, &listpayments_command);