Skip to contentSkip to conversation

Critical: Unbreak nfnl_osf tool

PATCH In review Phil Sutter 2 participants 5 messages May 15, 2020 – May 18, 2020 netfilter

lore
Share & tools
lore

Phil Sutter

author reviewer
#0 lore raw

v2 cover Critical: Unbreak nfnl_osf tool

No changes in patch 1, it is still the right fix for the problem and restores original behaviour.

Patch 2 changed according to feedback:

- Elaborate on why there are duplicates in pf.os in the first place.

- Ignore ENOENT when deleting. Since the code ignores EEXIST when creating, reporting this was asymmetrical behaviour.

- Fix for ugly error message when user didn't specify '-f' option.

Phil Sutter (2):
nfnl_osf: Fix broken conversion to nfnl_query()
nfnl_osf: Improve error handling

 utils/nfnl_osf.c | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)
Signature · 1 line
2.26.2
author reviewer
#1 lore raw

PATCH v2 1/2 nfnl_osf: Fix broken conversion to nfnl_query()

Due to missing NLM_F_ACK flag in request, nfnetlink code in kernel didn't create an own ACK message but left it upon subsystem to ACK or not. Since nfnetlink_osf doesn't ACK by itself, nfnl_query() got stuck waiting for a reply.

Whoever did the conversion from deprecated nfnl_talk() obviously didn't even test basic functionality of the tool.

Fixes 52aa15098ebd6 ("nfnl_osf: Replace deprecated nfnl_talk() by nfnl_query()")
---
 utils/nfnl_osf.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
utils/nfnl_osf.c +4 -2
index 15d531975e11d..922d90ac135b7 100644
@@ -378,9 +378,11 @@ static int osf_load_line(char *buffer, int len, int del)
 	memset(buf, 0, sizeof(buf));
 
 	if (del)
-		nfnl_fill_hdr(nfnlssh, nmh, 0, AF_UNSPEC, 0, OSF_MSG_REMOVE, NLM_F_REQUEST);
+		nfnl_fill_hdr(nfnlssh, nmh, 0, AF_UNSPEC, 0, OSF_MSG_REMOVE,
+			      NLM_F_ACK | NLM_F_REQUEST);
 	else
-		nfnl_fill_hdr(nfnlssh, nmh, 0, AF_UNSPEC, 0, OSF_MSG_ADD, NLM_F_REQUEST | NLM_F_CREATE);
+		nfnl_fill_hdr(nfnlssh, nmh, 0, AF_UNSPEC, 0, OSF_MSG_ADD,
+			      NLM_F_ACK | NLM_F_REQUEST | NLM_F_CREATE);
 
 	nfnl_addattr_l(nmh, sizeof(buf), OSF_ATTR_FINGER, &f, sizeof(struct xt_osf_user_finger));
 
Signature · 1 line
2.26.2
author reviewer
#2 lore raw

PATCH v2 2/2 nfnl_osf: Improve error handling

In reply to Phil Sutter

For some error cases, no log message was created - hence apart from the return code there was no indication of failing execution.

If a line load fails, don't abort but continue with the remaining file contents. The current pf.os file in this repository serves as proof-of-concept:

Lines 700 and 701: Duplicates of lines 698 and 699 because 'W*' and 'W0' parse into the same data.

Line 704: Duplicate of line 702 because apart from 'W*' and 'W0', only the first three fields on right-hand side are sent to the kernel.

When loading, these dups are ignored (they would bounce if NLM_F_EXCL was given). Upon deletion, they cause ENOENT response from kernel.

---
Changes since v1:
- Don't use ulog_err() when complaining about missing fingerprints argument, the use of strerror() with zero errno is misleading.
- Don't print error on osf_load_line() failure when deleting and errno is ENOENT. Upon add, NLM_F_EXCL is not set so EEXIST is basically ignored, be equally error-tolerant upon deletion.

---
 utils/nfnl_osf.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
utils/nfnl_osf.c +10 -5
index 922d90ac135b7..4183d66c22d16 100644
@@ -392,7 +392,7 @@ static int osf_load_line(char *buffer, int len, int del)
 static int osf_load_entries(char *path, int del)
 {
 	FILE *inf;
-	int err = 0;
+	int err = 0, lineno = 0;
 	char buf[1024];
 
 	inf = fopen(path, "r");
@@ -402,7 +402,9 @@ static int osf_load_entries(char *path, int del)
 	}
 
 	while(fgets(buf, sizeof(buf), inf)) {
-		int len;
+		int len, rc;
+
+		lineno++;
 
 		if (buf[0] == '#' || buf[0] == '\n' || buf[0] == '\r')
 			continue;
@@ -414,9 +416,11 @@ static int osf_load_entries(char *path, int del)
 
 		buf[len] = '\0';
 
-		err = osf_load_line(buf, len, del);
-		if (err)
-			break;
+		rc = osf_load_line(buf, len, del);
+		if (rc && (!del || errno == ENOENT)) {
+			ulog_err("Failed to load line %d", lineno);
+			err = rc;
+		}
 
 		memset(buf, 0, sizeof(buf));
 	}
@@ -448,6 +452,7 @@ int main(int argc, char *argv[])
 
 	if (!fingerprints) {
 		err = -ENOENT;
+		ulog("Missing fingerprints file argument.\n");
 		goto err_out_exit;
 	}
 
Signature · 1 line
2.26.2
author reviewer
#3 lore raw

Re: [iptables PATCH v2 2/2] nfnl_osf: Improve error handling

Two minor nits during final review:

On Fri, May 15, 2020 at 04:03:30PM +0200, Phil Sutter wrote:
[...]

Quoted 7 lines

---
Changes since v1:
- Don't use ulog_err() when complaining about missing fingerprints argument, the use of strerror() with zero errno is misleading.
- Don't print error on osf_load_line() failure when deleting and errno is ENOENT. Upon add, NLM_F_EXCL is not set so EEXIST is basically ignored, be equally error-tolerant upon deletion.

This change didn't make it into the commit message, although it's worth mentioning.

[...]

utils/nfnl_osf.c static int osf_load_entries(char *path, int del) · line 420

Quoted 9 lines
utils/nfnl_osf.c +2 -3
@@ -414,9 +416,11 @@ static int osf_load_entries(char *path, int del)
 
 		buf[len] = '\0';
 
-		err = osf_load_line(buf, len, del);
-		if (err)
-			break;
+		rc = osf_load_line(buf, len, del);
+		if (rc && (!del || errno == ENOENT)) {

Stupid typo here, it should read 'errno != ENOENT'.

I'll fix both before pushing upstream.

Cheers, Phil

#4 lore raw
In reply to Phil Sutter
Quoted 11 lines · Phil Sutter

No changes in patch 1, it is still the right fix for the problem and restores original behaviour.

Patch 2 changed according to feedback:

- Elaborate on why there are duplicates in pf.os in the first place.

- Ignore ENOENT when deleting. Since the code ignores EEXIST when creating, reporting this was asymmetrical behaviour.

- Fix for ugly error message when user didn't specify '-f' option.

This looks good.

Thanks for addressing my feedback.