Skip to contentSkip to conversation

ebtables: Dump atomic waste

PATCH In review Phil Sutter 2 participants 8 messages Jul 30, 2021 – Aug 2, 2021 netfilter

lore
Share & tools
lore

Phil Sutter

author reviewer
#0 lore raw

PATCH 1/1 ebtables: Dump atomic waste

With ebtables-nft.8 now educating people about the missing functionality, get rid of atomic remains in source code. This eliminates mostly comments except for --atomic-commit which was treated as alias of --init-table. People not using the latter are probably trying to atomic-commit from an atomic-file which in turn is not supported, so no point keeping it.

---
 iptables/xtables-eb.c | 53 -------------------------------------------
 1 file changed, 53 deletions(-)
iptables/xtables-eb.c +0 -53
index 6c58adaa66c1e..6e35f58ee685f 100644
@@ -211,10 +211,6 @@ struct option ebt_original_options[] =
 	{ "new-chain"      , required_argument, 0, 'N' },
 	{ "rename-chain"   , required_argument, 0, 'E' },
 	{ "delete-chain"   , optional_argument, 0, 'X' },
-	{ "atomic-init"    , no_argument      , 0, 7   },
-	{ "atomic-commit"  , no_argument      , 0, 8   },
-	{ "atomic-file"    , required_argument, 0, 9   },
-	{ "atomic-save"    , no_argument      , 0, 10  },
 	{ "init-table"     , no_argument      , 0, 11  },
 	{ "concurrent"     , no_argument      , 0, 13  },
 	{ 0 }
@@ -320,10 +316,6 @@ static void print_help(const struct xtables_target *t,
 "--new-chain -N chain          : create a user defined chain\n"
 "--rename-chain -E old new     : rename a chain\n"
 "--delete-chain -X [chain]     : delete a user defined chain\n"
-"--atomic-commit               : update the kernel w/t table contained in <FILE>\n"
-"--atomic-init                 : put the initial kernel table into <FILE>\n"
-"--atomic-save                 : put the current kernel table into <FILE>\n"
-"--atomic-file file            : set <FILE> to file\n\n"
 "Options:\n"
 "--proto  -p [!] proto         : protocol hexadecimal, by name or LENGTH\n"
 "--src    -s [!] address[/mask]: source mac address\n"
@@ -1087,54 +1079,9 @@ print_zero:
 					       "Use --Lmac2 with -L");
 			flags |= LIST_MAC2;
 			break;
-		case 8 : /* atomic-commit */
-/*
-			replace->command = c;
-			if (OPT_COMMANDS)
-				ebt_print_error2("Multiple commands are not allowed");
-			replace->flags |= OPT_COMMAND;
-			if (!replace->filename)
-				ebt_print_error2("No atomic file specified");*/
-			/* Get the information from the file */
-			/*ebt_get_table(replace, 0);*/
-			/* We don't want the kernel giving us its counters,
-			 * they would overwrite the counters extracted from
-			 * the file */
-			/*replace->num_counters = 0;*/
-			/* Make sure the table will be written to the kernel */
-			/*free(replace->filename);
-			replace->filename = NULL;
-			break;*/
-		/*case 7 :*/ /* atomic-init */
-		/*case 10:*/ /* atomic-save */
 		case 11: /* init-table */
 			nft_cmd_table_flush(h, *table, false);
 			return 1;
-		/*
-			replace->command = c;
-			if (OPT_COMMANDS)
-				ebt_print_error2("Multiple commands are not allowed");
-			if (c != 11 && !replace->filename)
-				ebt_print_error2("No atomic file specified");
-			replace->flags |= OPT_COMMAND;
-			{
-				char *tmp = replace->filename;*/
-
-				/* Get the kernel table */
-				/*replace->filename = NULL;
-				ebt_get_kernel_table(replace, c == 10 ? 0 : 1);
-				replace->filename = tmp;
-			}
-			break;
-		case 9 :*/ /* atomic */
-			/*
-			if (OPT_COMMANDS)
-				ebt_print_error2("--atomic has to come before the command");*/
-			/* A possible memory leak here, but this is not
-			 * executed in daemon mode */
-			/*replace->filename = (char *)malloc(strlen(optarg) + 1);
-			strcpy(replace->filename, optarg);
-			break; */
 		case 13 :
 			break;
 		case 1 :
Signature · 1 line
2.32.0
#1 lore raw

Hi Phil,

Quoted 6 lines · Phil Sutter

With ebtables-nft.8 now educating people about the missing functionality, get rid of atomic remains in source code. This eliminates mostly comments except for --atomic-commit which was treated as alias of --init-table. People not using the latter are probably trying to atomic-commit from an atomic-file which in turn is not supported, so no point keeping it.

That's fine.

If there's any need in the future for emulating this in the future, it should be possible to map atomic-save to ebtables-save and atomic-commit to ebtables-restore.

Anyway, this one of the exotic options in ebtables that makes it different from ip,ip6,arptables. Given there are better tools now that are aligned with the more orthodox approach, this should be OK.

Thanks.

author reviewer
#2 lore raw
Quoted 10 lines · Pablo Neira Ayuso

Hi Phil,

Quoted 6 lines · Phil Sutter

With ebtables-nft.8 now educating people about the missing functionality, get rid of atomic remains in source code. This eliminates mostly comments except for --atomic-commit which was treated as alias of --init-table. People not using the latter are probably trying to atomic-commit from an atomic-file which in turn is not supported, so no point keeping it.

Quoted 5 lines

That's fine.

If there's any need in the future for emulating this in the future, it should be possible to map atomic-save to ebtables-save and atomic-commit to ebtables-restore.

I had considered that, but the binary format of atomic-file drove me
off: If we can't support existing atomic-files easily, we better deny unless someone has a strong argument to do it. And then I'd try to support it fully, so it's not a half-ass solution with a catch. :)

Quoted 3 lines

Anyway, this one of the exotic options in ebtables that makes it different from ip,ip6,arptables. Given there are better tools now that are aligned with the more orthodox approach, this should be OK.

Let's hope most users went with the familiar save/restore approach instead of opening a whole new can for ebtables alone.

Thanks, Phil

#3 lore raw
Quoted 13 lines · Phil Sutter
Quoted 12 lines · Pablo Neira Ayuso

Hi Phil,

Quoted 6 lines · Phil Sutter

With ebtables-nft.8 now educating people about the missing functionality, get rid of atomic remains in source code. This eliminates mostly comments except for --atomic-commit which was treated as alias of --init-table. People not using the latter are probably trying to atomic-commit from an atomic-file which in turn is not supported, so no point keeping it.

That's fine.

Quoted 8 lines
Quoted 3 lines

If there's any need in the future for emulating this in the future, it should be possible to map atomic-save to ebtables-save and atomic-commit to ebtables-restore.

I had considered that, but the binary format of atomic-file drove me
off: If we can't support existing atomic-files easily, we better deny unless someone has a strong argument to do it. And then I'd try to support it fully, so it's not a half-ass solution with a catch. :)

That's sensible.

Quoted 6 lines
Quoted 3 lines

Anyway, this one of the exotic options in ebtables that makes it different from ip,ip6,arptables. Given there are better tools now that are aligned with the more orthodox approach, this should be OK.

Let's hope most users went with the familiar save/restore approach instead of opening a whole new can for ebtables alone.

Let's do that.

Thanks.

#4 lore raw

offlist

Quoted 18 lines · Pablo Neira Ayuso
Quoted 17 lines · Phil Sutter
Quoted 15 lines · Pablo Neira Ayuso

Hi Phil,

Quoted 6 lines · Phil Sutter

With ebtables-nft.8 now educating people about the missing functionality, get rid of atomic remains in source code. This eliminates mostly comments except for --atomic-commit which was treated as alias of --init-table. People not using the latter are probably trying to atomic-commit from an atomic-file which in turn is not supported, so no point keeping it.

That's fine.

If there's any need in the future for emulating this in the future, it should be possible to map atomic-save to ebtables-save and atomic-commit to ebtables-restore.

Quoted 6 lines
Quoted 4 lines

I had considered that, but the binary format of atomic-file drove me
off: If we can't support existing atomic-files easily, we better deny unless someone has a strong argument to do it. And then I'd try to support it fully, so it's not a half-ass solution with a catch. :)

That's sensible.

I did not know the file format is different, I never used exotic ebtables features ever myself.

author reviewer
#6 lore raw
Quoted 4 lines · Pablo Neira Ayuso
Quoted 1 line · Pablo Neira Ayuso

offlist

actually not :)

:D

The file is written by store_table_in_file() in communication.c, writing a struct ebt_replace plus a number of struct ebt_entries. At this point I stopped, assuming that translating back and forth would involve too much legacy cruft.

Cheers, Phil

#7 lore raw
Quoted 4 lines · Phil Sutter
Quoted 3 lines · Pablo Neira Ayuso
Quoted 1 line · Pablo Neira Ayuso

offlist

Quoted 8 lines
Quoted 1 line

actually not :)

:D

The file is written by store_table_in_file() in communication.c, writing a struct ebt_replace plus a number of struct ebt_entries. At this point I stopped, assuming that translating back and forth would involve too much legacy cruft.

Hm, so it looks like a memory dump of the ruleset blob representation into a file.