Skip to contentSkip to conversation

virtiofsd xattr name mappings

PATCH Superseded Dr. David Alan Gilbert (git) 5 participants 32 messages Oct 14, 2020 – Oct 23, 2020

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover virtiofsd xattr name mappings

From: "Dr. David Alan Gilbert" <[email protected]>

This is a 3rd cut of an xattr name mapping option for virtiofsd. It allows the user of virtiofsd to define a fairly flexible mapping from the view of the xattr names the host fs has and the ones that the guest sees.

The hope is this allows things like:
a) Different selinux attributes on host/guest
b) separation of trusted. attributes that clash on overlayfs
c) support for privileged xattr's in guests running with an unprivileged virtiofsd.

There's no apparent standard for this kind of mapping, so I made it flexible by specifying a mapping rule in the option.

Prefix's can be added (selectively or globally), xattr's can be dropped in either direction or passed through.

The major change for v3 is the addition of a 'map' simple syntax that should cover a lot of the simple cases without people needing to use the more complex rule syntax. There's also some cleanups basedon reviews by Christophe.

Dave

Dr. David Alan Gilbert (5):
tools/virtiofsd: xattr name mappings: Add option tools/virtiofsd: xattr name mappings: Map client xattr names tools/virtiofsd: xattr name mappings: Map server xattr names tools/virtiofsd: xattr name mapping examples tools/virtiofsd: xattr name mappings: Simple 'map'

 docs/tools/virtiofsd.rst         | 122 ++++++++
 tools/virtiofsd/passthrough_ll.c | 476 ++++++++++++++++++++++++++++++-
 2 files changed, 595 insertions(+), 3 deletions(-)
Signature · 1 line
2.28.0
#1 lore raw

PATCH v3 1/5 tools/virtiofsd: xattr name mappings: Add option

From: "Dr. David Alan Gilbert" <[email protected]>

Add an option to define mappings of xattr names so that the client and server filesystems see different views. This can be used to have different SELinux mappings as seen by the guest, to run the virtiofsd with less privileges (e.g. in a case where it can't set trusted/system/security xattrs but you want the guest to be able to), or to isolate multiple users of the same name; e.g. trusted attributes used by stacking overlayfs.

A mapping engine is used wit 3 simple rules; the rules can be combined to allow most useful mapping scenarios. The ruleset is defined by -o xattrmap='rules...'.

This patch doesn't use the rule maps yet.

---
 docs/tools/virtiofsd.rst         |  54 ++++++++++
 tools/virtiofsd/passthrough_ll.c | 180 +++++++++++++++++++++++++++++++
 2 files changed, 234 insertions(+)
docs/tools/virtiofsd.rst +54 -0
index 7ecee49834..a3a120da2f 100644
@@ -109,6 +109,60 @@ Options
   timeout.  ``always`` sets a long cache lifetime at the expense of coherency.
   The default is ``auto``.
 
+xattr-mapping
+-------------
+
+By default the name of xattr's used by the client are passed through to the server
+file system.  This can be a problem where either those xattr names are used
+by something on the server (e.g. selinux client/server confusion) or if the
+virtiofsd is running in a container with restricted privileges where it cannot
+access some attributes.
+
+A mapping of xattr names can be made using -o xattrmap=mapping where the ``mapping``
+string consists of a series of rules.
+
+The first matching rule terminates the mapping.
+
+Each rule consists of a number of fields separated with a separator that is the
+first non-white space character in the rule.  This separator must then be used
+for the whole rule.
+White space may be added before and after each rule.
+Using ':' as the separator a rule is of the form:
+
+``:type:scope:key:prepend:``
+
+**type** is one of:
+
+- 'prefix' - If 'key' matches the client then the 'prepend'
+  is added before the name is passed to the server.
+  For a server case, the prepend is tested and stripped
+  if matching.
+
+- 'ok' - The attribute name is OK and passed through to
+  the server unchanged.
+
+- 'bad' - If a client tries to use this name it's
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
+
+**scope** is:
+
+- 'client' - match 'key' against a xattr name from the client for
+             setxattr/getxattr/removexattr
+- 'server' - match 'prepend' against a xattr name from the server
+             for listxattr
+- 'all' - can be used to match both cases.
+
+**key** is a string tested as a prefix on an attribute name originating
+on the client.  It maybe empty in which case a 'client' rule
+will always match on client names.
+
+**prepend** is a string tested as a prefix on an attribute name originating
+on the server, and used as a new prefix.  It may be empty
+in which case a 'server' rule will always match on all names from
+the server.
+
+
 Examples
 --------
 
tools/virtiofsd/passthrough_ll.c +180 -0

228 lines not on this page · raw message

Signature · 1 line
2.28.0
#2 lore raw

Re: [PATCH v3 1/5] tools/virtiofsd: xattr name mappings: Add option

Quoted 11 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add an option to define mappings of xattr names so that the client and server filesystems see different views. This can be used to have different SELinux mappings as seen by the guest, to run the virtiofsd with less privileges (e.g. in a case where it can't set trusted/system/security xattrs but you want the guest to be able to), or to isolate multiple users of the same name; e.g. trusted attributes used by stacking overlayfs.

Quoted 1 line

A mapping engine is used wit 3 simple rules; the rules can

s/wit/with/

docs/tools/virtiofsd.rst Options · line 139

Quoted 8 lines

+``:type:scope:key:prepend:``

docs/tools/virtiofsd.rst +7 -0
+
+**type** is one of:
+
+- 'prefix' - If 'key' matches the client then the 'prepend'
+  is added before the name is passed to the server.
+  For a server case, the prepend is tested and stripped
+  if matching.

It may be clearer to document rule types like this:

- :prefix:client:key:prepend: - Add 'prepend' before the name if it starts with 'key'.

- :prefix:server::prepend: - Strip 'prepend' if the name starts with

                               it.

The client vs server behavior is independent so it's clearer to list them as separate cases. In addition, using the full rule syntax shows which fields are valid arguments and which ones are ignored.

The 'all' scope can be documented later as "Combines both the 'client' and 'server' scope behavior".

docs/tools/virtiofsd.rst Options · line 142

Quoted 3 lines
docs/tools/virtiofsd.rst +3 -0
+
+- 'ok' - The attribute name is OK and passed through to
+  the server unchanged.

The documentation isn't explicit but I think the default behavior is to allow all xattr names?

What is the purpose of the 'ok' rule? I guess it's to define an exception to a later 'prefix' or 'bad' rule. It would be nice to make this clear.

The documentation only mentions :client: behavior, leaving :server:
undefined. Please indicate whether this rule has an effect in server scope.

Quoted 6 lines
docs/tools/virtiofsd.rst +6 -0
+
+- 'bad' - If a client tries to use this name it's
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
+
+**scope** is:

docs/tools/virtiofsd.rst Options · line 158

Quoted 10 lines
docs/tools/virtiofsd.rst +10 -0
+
+- 'client' - match 'key' against a xattr name from the client for
+             setxattr/getxattr/removexattr
+- 'server' - match 'prepend' against a xattr name from the server
+             for listxattr
+- 'all' - can be used to match both cases.
+
+**key** is a string tested as a prefix on an attribute name originating
+on the client.  It maybe empty in which case a 'client' rule
+will always match on client names.

Is there a way to match a full string instead of a prefix (regexp ^<pattern>$ instead of ^<pattern>)?

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2030

Quoted 12 lines
tools/virtiofsd/passthrough_ll.c +8 -0
@@ -2010,6 +2020,169 @@ static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi,
     fuse_reply_err(req, res == -1 ? errno : 0);
 }
 
+/*
+ * Exit; process attribute unmodified if matched.
+ * An empty key applies to all.
+ */
+#define XATTR_MAP_FLAG_END_OK  (1 <<  0)
+/*
+ * The attribute is unwanted;
+ * EPERM on write hidden on read.

Making this sentence easier to parse:

s/write hidden/write, hidden/

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2035

Quoted 5 lines
tools/virtiofsd/passthrough_ll.c +5 -0
+ */
+#define XATTR_MAP_FLAG_END_BAD (1 <<  1)
+/*
+ * For attr that start with 'key' prepend 'prepend'
+ * 'key' maybe empty to prepend for all attrs

s/maybe/may be/

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2177

Quoted 3 lines
tools/virtiofsd/passthrough_ll.c +3 -0
+    /* Add a terminator to error in cases the user hasn't specified */
+    tmp_entry.flags = XATTR_MAP_FLAG_ALL | XATTR_MAP_FLAG_END_BAD |
+                      XATTR_MAP_FLAG_LAST;

The comment is slightly misleading. This entry must be added in all cases since it terminates the lo->xattr_map_list with the XATTR_MAP_FLAG_LAST flag. If we don't add this entry then free_xattrmap() will iterate beyond the end of lo->xattr_map_list.

Another approach is to set XATTR_MAP_FLAG_LAST in add_xattrmap_entry() (and clear it on the previous last entry). That way adding the 'bad' catch-all truly is optional and just for cases where the user hasn't defined a catch-all rule themselves.

Quoted 21 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+    tmp_entry.key = g_strdup("");
+    tmp_entry.prepend = g_strdup("");
+    lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                            &tmp_entry);
+
+    return res;
+}
+
 static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
                         size_t size)
 {
@@ -2806,6 +2979,8 @@ static void fuse_lo_data_cleanup(struct lo_data *lo)
         close(lo->root.fd);
     }
 
+    free(lo->xattrmap);
+    free_xattrmap(lo->xattr_map_list);
     free(lo->source);
 }
 

tools/virtiofsd/passthrough_ll.c int main(int argc, char *argv[]) · line 3087

Quoted 7 lines
tools/virtiofsd/passthrough_ll.c +4 -0
@@ -2906,6 +3081,11 @@ int main(int argc, char *argv[])
     } else {
         lo.source = strdup("/");
     }
+
+    if (lo.xattrmap) {
+        lo.xattr_map_list = parse_xattrmap(&lo);
+    }

The function always returns NULL. Has this been tested?

#3 lore raw

Re: [PATCH v3 1/5] tools/virtiofsd: xattr name mappings: Add option

Quoted 12 lines · * Stefan Hajnoczi ([email protected]) wrote:
Quoted 11 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add an option to define mappings of xattr names so that the client and server filesystems see different views. This can be used to have different SELinux mappings as seen by the guest, to run the virtiofsd with less privileges (e.g. in a case where it can't set trusted/system/security xattrs but you want the guest to be able to), or to isolate multiple users of the same name; e.g. trusted attributes used by stacking overlayfs.

Quoted 3 lines
Quoted 1 line

A mapping engine is used wit 3 simple rules; the rules can

s/wit/with/

Done.

Quoted 15 lines

docs/tools/virtiofsd.rst Options · line 139

Quoted 8 lines

+``:type:scope:key:prepend:``

docs/tools/virtiofsd.rst +7 -0
+
+**type** is one of:
+
+- 'prefix' - If 'key' matches the client then the 'prepend'
+  is added before the name is passed to the server.
+  For a server case, the prepend is tested and stripped
+  if matching.

It may be clearer to document rule types like this:

- :prefix:client:key:prepend: - Add 'prepend' before the name if it starts with 'key'.

- :prefix:server::prepend: - Strip 'prepend' if the name starts with

Quoted 8 lines
                               it.

The client vs server behavior is independent so it's clearer to list them as separate cases. In addition, using the full rule syntax shows which fields are valid arguments and which ones are ignored.

The 'all' scope can be documented later as "Combines both the 'client' and 'server' scope behavior".

OK, I've reworked this quite a bit, into a simpler part for each of the type entries with examples of each below.

Quoted 7 lines

docs/tools/virtiofsd.rst Options · line 142

Quoted 3 lines
docs/tools/virtiofsd.rst +3 -0
+
+- 'ok' - The attribute name is OK and passed through to
+  the server unchanged.

The documentation isn't explicit but I think the default behavior is to allow all xattr names?

Quoted 7 lines

What is the purpose of the 'ok' rule? I guess it's to define an exception to a later 'prefix' or 'bad' rule. It would be nice to make this clear.

The documentation only mentions :client: behavior, leaving :server:
undefined. Please indicate whether this rule has an effect in server scope.

What I have now is:
+**scope** is:
+

+- 'client' - match 'key' against a xattr name from the client for
+             setxattr/getxattr/removexattr
+- 'server' - match 'prepend' against a xattr name from the server
+             for listxattr
+- 'all' - can be used to make a single rule where both the server
+          and client matches are triggered.
+
+**type** is one of:
+
+- 'prefix' - is designed to prepend and strip a prefix;  the modified
+  attributes then being passed on to the client/server.
+
+- 'ok' - Causes the rule set to be terminated when a match is found
+  while allowing matching xattr's through unchanged.
+  It is intended both as a way of explicitly terminating
+  the list of rules, and to allow some xattr's to skip following rules.
+
+- 'bad' - If a client tries to use a name matching 'key' it's
+  denied using EPERM; when the server passes an attribute
+  name matching 'prepend' it's hidden.  In many ways it's use is very like
+  'ok' as either an explict terminator or for special handling of certain
+  patterns.

+
+**key** is a string tested as a prefix on an attribute name originating +on the client. It maybe empty in which case a 'client' rule +will always match on client names.
+
+**prepend** is a string tested as a prefix on an attribute name originating +on the server, and used as a new prefix. It may be empty
+in which case a 'server' rule will always match on all names from +the server.
+
+e.g.:
+
+ ``:prefix:client:trusted.:user.virtiofs.:``
+
+ will match 'trusted.' attributes in client calls and prefix them before
+ passing them to the server.
+
+ ``:prefix:server::user.virtiofs.:``
+
+ will strip 'user.virtiofs.' from all server replies.
+
+ ``:prefix:all:trusted.:user.virtiofs.:``
+
+ combines the previous two cases into a single rule.
+
+ ``:ok:client:user.::``
+
+ will allow get/set xattr for 'user.' xattr's and ignore
+ following rules.
+
+ ``:ok:server::security.:``
+
+ will pass 'securty.' xattr's in listxattr from the server
+ and ignore following rules.
+
+ ``:ok:all:::``
+
+ will terminate the rule search passing any remaining attributes
+ in both directions.
+
+ ``:bad:server::security.:``
+
+ would hide 'security.' xattr's in listxattr from the server.

so I'm hoping that addresses both the prefix and OK sections at least.

Quoted 11 lines
Quoted 11 lines
docs/tools/virtiofsd.rst +11 -0
+
+- 'bad' - If a client tries to use this name it's
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
+
+**scope** is:
+
+- 'client' - match 'key' against a xattr name from the client for
+             setxattr/getxattr/removexattr
+- 'server' - match 'prepend' against a xattr name from the server
+             for listxattr
Quoted 8 lines

docs/tools/virtiofsd.rst Options · line 158

Quoted 5 lines

+- 'all' - can be used to match both cases.

docs/tools/virtiofsd.rst +4 -0
+
+**key** is a string tested as a prefix on an attribute name originating
+on the client.  It maybe empty in which case a 'client' rule
+will always match on client names.

Is there a way to match a full string instead of a prefix (regexp ^<pattern>$ instead of ^<pattern>)?

No there isn't; can you think of a way of representing that in the syntax without making it much more complex?

Quoted 15 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2030

Quoted 12 lines
tools/virtiofsd/passthrough_ll.c +8 -0
@@ -2010,6 +2020,169 @@ static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi,
     fuse_reply_err(req, res == -1 ? errno : 0);
 }
 
+/*
+ * Exit; process attribute unmodified if matched.
+ * An empty key applies to all.
+ */
+#define XATTR_MAP_FLAG_END_OK  (1 <<  0)
+/*
+ * The attribute is unwanted;
+ * EPERM on write hidden on read.

Making this sentence easier to parse:

Quoted 1 line

s/write hidden/write, hidden/

Done.

Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2035

Quoted 6 lines
tools/virtiofsd/passthrough_ll.c +5 -0
+ */
+#define XATTR_MAP_FLAG_END_BAD (1 <<  1)
+/*
+ * For attr that start with 'key' prepend 'prepend'
+ * 'key' maybe empty to prepend for all attrs

s/maybe/may be/

Hmm OK.

Quoted 9 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2177

Quoted 3 lines
tools/virtiofsd/passthrough_ll.c +3 -0
+    /* Add a terminator to error in cases the user hasn't specified */
+    tmp_entry.flags = XATTR_MAP_FLAG_ALL | XATTR_MAP_FLAG_END_BAD |
+                      XATTR_MAP_FLAG_LAST;

The comment is slightly misleading. This entry must be added in all cases since it terminates the lo->xattr_map_list with the XATTR_MAP_FLAG_LAST flag. If we don't add this entry then free_xattrmap() will iterate beyond the end of lo->xattr_map_list.

Quoted 4 lines

Another approach is to set XATTR_MAP_FLAG_LAST in add_xattrmap_entry() (and clear it on the previous last entry). That way adding the 'bad' catch-all truly is optional and just for cases where the user hasn't defined a catch-all rule themselves.

I've changed the comment.

Quoted 20 lines
Quoted 20 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+    tmp_entry.key = g_strdup("");
+    tmp_entry.prepend = g_strdup("");
+    lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                            &tmp_entry);
+
+    return res;
+}
+
 static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
                         size_t size)
 {
@@ -2806,6 +2979,8 @@ static void fuse_lo_data_cleanup(struct lo_data *lo)
         close(lo->root.fd);
     }
 
+    free(lo->xattrmap);
+    free_xattrmap(lo->xattr_map_list);
     free(lo->source);
 }
 
Quoted 10 lines

tools/virtiofsd/passthrough_ll.c int main(int argc, char *argv[]) · line 3087

Quoted 8 lines
tools/virtiofsd/passthrough_ll.c +4 -0
@@ -2906,6 +3081,11 @@ int main(int argc, char *argv[])
     } else {
         lo.source = strdup("/");
     }
+
+    if (lo.xattrmap) {
+        lo.xattr_map_list = parse_xattrmap(&lo);
+    }

The function always returns NULL. Has this been tested?

Hmm; I moved that xattr_map_list late and only retested with the 'map' shortcut which still returned it. Fixed.

Dave

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#4 lore raw

Re: [PATCH v3 1/5] tools/virtiofsd: xattr name mappings: Add option

In reply to Dr. David Alan Gilbert (git)
Quoted 38 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add an option to define mappings of xattr names so that the client and server filesystems see different views. This can be used to have different SELinux mappings as seen by the guest, to run the virtiofsd with less privileges (e.g. in a case where it can't set trusted/system/security xattrs but you want the guest to be able to), or to isolate multiple users of the same name; e.g. trusted attributes used by stacking overlayfs.

A mapping engine is used wit 3 simple rules; the rules can be combined to allow most useful mapping scenarios. The ruleset is defined by -o xattrmap='rules...'.

This patch doesn't use the rule maps yet.

---
 docs/tools/virtiofsd.rst         |  54 ++++++++++
 tools/virtiofsd/passthrough_ll.c | 180 +++++++++++++++++++++++++++++++
 2 files changed, 234 insertions(+)
docs/tools/virtiofsd.rst +7 -0
index 7ecee49834..a3a120da2f 100644
@@ -109,6 +109,60 @@ Options
   timeout.  ``always`` sets a long cache lifetime at the expense of coherency.
   The default is ``auto``.
 
+xattr-mapping
+-------------
+
+By default the name of xattr's used by the client are passed through to the server
+file system.  This can be a problem where either those xattr names are used
+by something on the server (e.g. selinux client/server confusion) or if the
+virtiofsd is running in a container with restricted privileges where it cannot

docs/tools/virtiofsd.rst Options · line 128

Quoted 10 lines
docs/tools/virtiofsd.rst +10 -0
+access some attributes.
+
+A mapping of xattr names can be made using -o xattrmap=mapping where the ``mapping``
+string consists of a series of rules.
+
+The first matching rule terminates the mapping.
+
+Each rule consists of a number of fields separated with a separator that is the
+first non-white space character in the rule.  This separator must then be used
+for the whole rule.

Where is it useful to have a separator other than ":".

docs/tools/virtiofsd.rst Options · line 134

Quoted 6 lines
docs/tools/virtiofsd.rst +6 -0
+White space may be added before and after each rule.
+Using ':' as the separator a rule is of the form:
+
+``:type:scope:key:prepend:``
+
+**type** is one of:

I am not sure I understand the syntax explanation. So I will ask some basic questions.

docs/tools/virtiofsd.rst Options · line 136

Quoted 2 lines
docs/tools/virtiofsd.rst +2 -0
+
+- 'prefix' - If 'key' matches the client then the 'prepend'

What do you mean by "If key matches the client"?

docs/tools/virtiofsd.rst Options · line 137

Quoted 1 line
docs/tools/virtiofsd.rst +1 -0
+  is added before the name is passed to the server.

Its basically the sever which is processing these rules. I guess you have written this sentence from the perspective of "rule engine" and it passes the name to file server (despite the fact that rule engine itself is part of file server).

docs/tools/virtiofsd.rst Options · line 139

Quoted 2 lines
docs/tools/virtiofsd.rst +2 -0
+  For a server case, the prepend is tested and stripped
+  if matching.

Is this about file server removing the "prepend" from xattr name before it is sent back to client?

Quoted 5 lines
docs/tools/virtiofsd.rst +5 -0
+
+- 'ok' - The attribute name is OK and passed through to
+  the server unchanged.
+
+- 'bad' - If a client tries to use this name it's

docs/tools/virtiofsd.rst Options · line 154

Quoted 10 lines
docs/tools/virtiofsd.rst +10 -0
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
+
+**scope** is:
+
+- 'client' - match 'key' against a xattr name from the client for
+             setxattr/getxattr/removexattr
+- 'server' - match 'prepend' against a xattr name from the server
+             for listxattr
+- 'all' - can be used to match both cases.

So we need this only if we don't want to do xattr mapping bidirectionally. IOW, server will not allow getxattr "user.foo" but will allow listing "user.foo"?

/me is wondering what's the use case of this.

docs/tools/virtiofsd.rst Options

Quoted 7 lines
docs/tools/virtiofsd.rst +7 -0
+
+**key** is a string tested as a prefix on an attribute name originating
+on the client.  It maybe empty in which case a 'client' rule
+will always match on client names.
+
+**prepend** is a string tested as a prefix on an attribute name originating
+on the server, and used as a new prefix.

What does it mean that "prepend" is tested as a prefix" and "used as a new prefix". It is used as a new prefix for client rule, right? Atleast documentation does not make it clear.

docs/tools/virtiofsd.rst Options · line 167

Quoted 7 lines

It may be empty

docs/tools/virtiofsd.rst +4 -0
+in which case a 'server' rule will always match on all names from
+the server.
+
+
 Examples
 --------

Can we give some examples in this patch so that it is easy to understand what these rules can allow.

Thanks
Vivek

Quoted 244 lines
tools/virtiofsd/passthrough_ll.c +180 -0
index ff53df4451..f5a33014f9 100644
@@ -64,6 +64,7 @@
 #include <syslog.h>
 #include <unistd.h>
 
+#include "qemu/cutils.h"
 #include "passthrough_helpers.h"
 #include "passthrough_seccomp.h"
 
@@ -137,6 +138,12 @@ enum {
     CACHE_ALWAYS,
 };
 
+typedef struct xattr_map_entry {
+    char *key;
+    char *prepend;
+    unsigned int flags;
+} XattrMapEntry;
+
 struct lo_data {
     pthread_mutex_t mutex;
     int debug;
@@ -144,6 +151,7 @@ struct lo_data {
     int flock;
     int posix_lock;
     int xattr;
+    char *xattrmap;
     char *source;
     char *modcaps;
     double timeout;
@@ -157,6 +165,7 @@ struct lo_data {
     struct lo_map ino_map; /* protected by lo->mutex */
     struct lo_map dirp_map; /* protected by lo->mutex */
     struct lo_map fd_map; /* protected by lo->mutex */
+    XattrMapEntry *xattr_map_list;
 
     /* An O_PATH file descriptor to /proc/self/fd/ */
     int proc_self_fd;
@@ -172,6 +181,7 @@ static const struct fuse_opt lo_opts[] = {
     { "no_posix_lock", offsetof(struct lo_data, posix_lock), 0 },
     { "xattr", offsetof(struct lo_data, xattr), 1 },
     { "no_xattr", offsetof(struct lo_data, xattr), 0 },
+    { "xattrmap=%s", offsetof(struct lo_data, xattrmap), 0 },
     { "modcaps=%s", offsetof(struct lo_data, modcaps), 0 },
     { "timeout=%lf", offsetof(struct lo_data, timeout), 0 },
     { "timeout=", offsetof(struct lo_data, timeout_set), 1 },
@@ -2010,6 +2020,169 @@ static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi,
     fuse_reply_err(req, res == -1 ? errno : 0);
 }
 
+/*
+ * Exit; process attribute unmodified if matched.
+ * An empty key applies to all.
+ */
+#define XATTR_MAP_FLAG_END_OK  (1 <<  0)
+/*
+ * The attribute is unwanted;
+ * EPERM on write hidden on read.
+ */
+#define XATTR_MAP_FLAG_END_BAD (1 <<  1)
+/*
+ * For attr that start with 'key' prepend 'prepend'
+ * 'key' maybe empty to prepend for all attrs
+ * key is defined from set/remove point of view.
+ * Automatically reversed on read
+ */
+#define XATTR_MAP_FLAG_PREFIX  (1 <<  2)
+/* Apply rule to get/set/remove */
+#define XATTR_MAP_FLAG_CLIENT  (1 << 16)
+/* Apply rule to list */
+#define XATTR_MAP_FLAG_SERVER  (1 << 17)
+/* Apply rule to all */
+#define XATTR_MAP_FLAG_ALL   (XATTR_MAP_FLAG_SERVER | XATTR_MAP_FLAG_CLIENT)
+
+/* Last rule in the XATTR_MAP */
+#define XATTR_MAP_FLAG_LAST    (1 << 30)
+
+static XattrMapEntry *add_xattrmap_entry(XattrMapEntry *orig_map,
+                                         size_t *nentries,
+                                         const XattrMapEntry *new_entry)
+{
+    XattrMapEntry *res = g_realloc_n(orig_map, ++*nentries,
+                                     sizeof(XattrMapEntry));
+    res[*nentries - 1] = *new_entry;
+
+    return res;
+}
+
+static void free_xattrmap(XattrMapEntry *map)
+{
+    XattrMapEntry *curr = map;
+
+    if (!map) {
+        return;
+    };
+
+    do {
+        g_free(curr->key);
+        g_free(curr->prepend);
+    } while (!(curr++->flags & XATTR_MAP_FLAG_LAST));
+
+    g_free(map);
+}
+
+static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
+{
+    XattrMapEntry *res = NULL;
+    XattrMapEntry tmp_entry;
+    size_t nentries = 0;
+    const char *map = lo->xattrmap;
+    const char *tmp;
+
+    while (*map) {
+        char sep;
+
+        if (isspace(*map)) {
+            map++;
+            continue;
+        }
+        /* The separator is the first non-space of the rule */
+        sep = *map++;
+        if (!sep) {
+            break;
+        }
+
+        /* Start of 'type' */
+        if (strstart(map, "prefix", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_PREFIX;
+        } else if (strstart(map, "ok", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_OK;
+        } else if (strstart(map, "bad", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_BAD;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected type;"
+                     "Expecting 'prefix', 'ok', or 'bad' in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of type field of rule %zu\n",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+
+        /* Start of 'scope' */
+        if (strstart(map, "client", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_CLIENT;
+        } else if (strstart(map, "server", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_SERVER;
+        } else if (strstart(map, "all", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_ALL;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected scope;"
+                     " Expecting 'client', 'server', or 'all', in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Expecting '%c' found '%c'"
+                     " after scope in rule %zu\n",
+                     __func__, sep, *map, nentries);
+            exit(1);
+        }
+
+        /* At start of 'key' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of key field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.key = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        /* At start of 'prepend' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of prepend field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.prepend = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                                &tmp_entry);
+        /* End of rule - go around again for another rule */
+    }
+
+    if (!nentries) {
+        fuse_log(FUSE_LOG_ERR, "Empty xattr map\n");
+        exit(1);
+    }
+
+    /* Add a terminator to error in cases the user hasn't specified */
+    tmp_entry.flags = XATTR_MAP_FLAG_ALL | XATTR_MAP_FLAG_END_BAD |
+                      XATTR_MAP_FLAG_LAST;
+    tmp_entry.key = g_strdup("");
+    tmp_entry.prepend = g_strdup("");
+    lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                            &tmp_entry);
+
+    return res;
+}
+
 static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
                         size_t size)
 {
@@ -2806,6 +2979,8 @@ static void fuse_lo_data_cleanup(struct lo_data *lo)
         close(lo->root.fd);
     }
 
+    free(lo->xattrmap);
+    free_xattrmap(lo->xattr_map_list);
     free(lo->source);
 }
 
@@ -2906,6 +3081,11 @@ int main(int argc, char *argv[])
     } else {
         lo.source = strdup("/");
     }
+
+    if (lo.xattrmap) {
+        lo.xattr_map_list = parse_xattrmap(&lo);
+    }
+
     if (!lo.timeout_set) {
         switch (lo.cache) {
         case CACHE_NONE:
Signature · 1 line
2.28.0
#5 lore raw

Re: [PATCH v3 1/5] tools/virtiofsd: xattr name mappings: Add option

In reply to Dr. David Alan Gilbert (git)

On Wed, Oct 14, 2020 at 07:02:05PM +0100, Dr. David Alan Gilbert (git) wrote:

[..]

Quoted 16 lines
tools/virtiofsd/passthrough_ll.c +16 -0
+/*
+ * Exit; process attribute unmodified if matched.
+ * An empty key applies to all.
+ */
+#define XATTR_MAP_FLAG_END_OK  (1 <<  0)
+/*
+ * The attribute is unwanted;
+ * EPERM on write hidden on read.
+ */
+#define XATTR_MAP_FLAG_END_BAD (1 <<  1)
+/*
+ * For attr that start with 'key' prepend 'prepend'
+ * 'key' maybe empty to prepend for all attrs
+ * key is defined from set/remove point of view.
+ * Automatically reversed on read
+ */

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2048

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+#define XATTR_MAP_FLAG_PREFIX  (1 <<  2)
+/* Apply rule to get/set/remove */
+#define XATTR_MAP_FLAG_CLIENT  (1 << 16)
+/* Apply rule to list */
+#define XATTR_MAP_FLAG_SERVER  (1 << 17)
+/* Apply rule to all */
+#define XATTR_MAP_FLAG_ALL   (XATTR_MAP_FLAG_SERVER | XATTR_MAP_FLAG_CLIENT)
+
+/* Last rule in the XATTR_MAP */
+#define XATTR_MAP_FLAG_LAST    (1 << 30)

I see that you are using bit positions for flags. Not clear why you used bit 0,1,2 and then jumped to 16,17 and then to 30. May be you are doing some sort of reservation of bits. Will be nice to explain that a bit so that next person modifying it can use bits from correct pool.

Quoted 9 lines
tools/virtiofsd/passthrough_ll.c +9 -0
+
+static XattrMapEntry *add_xattrmap_entry(XattrMapEntry *orig_map,
+                                         size_t *nentries,
+                                         const XattrMapEntry *new_entry)
+{
+    XattrMapEntry *res = g_realloc_n(orig_map, ++*nentries,
+                                     sizeof(XattrMapEntry));
+    res[*nentries - 1] = *new_entry;
+

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2067

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+    return res;
+}
+
+static void free_xattrmap(XattrMapEntry *map)
+{
+    XattrMapEntry *curr = map;
+
+    if (!map) {
+        return;
+    };

; after } is not needed.

Quoted 4 lines
tools/virtiofsd/passthrough_ll.c +4 -0
+
+    do {
+        g_free(curr->key);
+        g_free(curr->prepend);

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2081

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+    } while (!(curr++->flags & XATTR_MAP_FLAG_LAST));
+
+    g_free(map);
+}
+
+static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
+{
+    XattrMapEntry *res = NULL;
+    XattrMapEntry tmp_entry;
+    size_t nentries = 0;

If you are calculating number of entries (nentries), may be this could be stored in lo_data so that can be later used to free entries or loop through rules etc.

Quoted 5 lines
tools/virtiofsd/passthrough_ll.c +5 -0
+    const char *map = lo->xattrmap;
+    const char *tmp;
+
+    while (*map) {
+        char sep;

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2096

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+
+        if (isspace(*map)) {
+            map++;
+            continue;
+        }
+        /* The separator is the first non-space of the rule */
+        sep = *map++;
+        if (!sep) {
+            break;
+        }

When can sep be NULL? In that case while loop will not even continue.

Quoted 75 lines
tools/virtiofsd/passthrough_ll.c +75 -0
+
+        /* Start of 'type' */
+        if (strstart(map, "prefix", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_PREFIX;
+        } else if (strstart(map, "ok", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_OK;
+        } else if (strstart(map, "bad", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_BAD;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected type;"
+                     "Expecting 'prefix', 'ok', or 'bad' in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of type field of rule %zu\n",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+
+        /* Start of 'scope' */
+        if (strstart(map, "client", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_CLIENT;
+        } else if (strstart(map, "server", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_SERVER;
+        } else if (strstart(map, "all", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_ALL;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected scope;"
+                     " Expecting 'client', 'server', or 'all', in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Expecting '%c' found '%c'"
+                     " after scope in rule %zu\n",
+                     __func__, sep, *map, nentries);
+            exit(1);
+        }
+
+        /* At start of 'key' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of key field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.key = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        /* At start of 'prepend' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of prepend field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.prepend = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                                &tmp_entry);
+        /* End of rule - go around again for another rule */
+    }
+
+    if (!nentries) {
+        fuse_log(FUSE_LOG_ERR, "Empty xattr map\n");

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2181

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+        exit(1);
+    }
+
+    /* Add a terminator to error in cases the user hasn't specified */
+    tmp_entry.flags = XATTR_MAP_FLAG_ALL | XATTR_MAP_FLAG_END_BAD |
+                      XATTR_MAP_FLAG_LAST;
+    tmp_entry.key = g_strdup("");
+    tmp_entry.prepend = g_strdup("");
+    lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                            &tmp_entry);

Not sure why this default rule is needed when user has not specified one.

This seems to be equivalent of ":bad:all:::". Will this not block all the xattrs which have not been caught by previous rules. And user probably did not want it.

Thanks
Vivek

#6 lore raw

Re: [PATCH v3 1/5] tools/virtiofsd: xattr name mappings: Add option

Quoted 27 lines · * Vivek Goyal ([email protected]) wrote:

On Wed, Oct 14, 2020 at 07:02:05PM +0100, Dr. David Alan Gilbert (git) wrote:

[..]

Quoted 24 lines
tools/virtiofsd/passthrough_ll.c +24 -0
+/*
+ * Exit; process attribute unmodified if matched.
+ * An empty key applies to all.
+ */
+#define XATTR_MAP_FLAG_END_OK  (1 <<  0)
+/*
+ * The attribute is unwanted;
+ * EPERM on write hidden on read.
+ */
+#define XATTR_MAP_FLAG_END_BAD (1 <<  1)
+/*
+ * For attr that start with 'key' prepend 'prepend'
+ * 'key' maybe empty to prepend for all attrs
+ * key is defined from set/remove point of view.
+ * Automatically reversed on read
+ */
+#define XATTR_MAP_FLAG_PREFIX  (1 <<  2)
+/* Apply rule to get/set/remove */
+#define XATTR_MAP_FLAG_CLIENT  (1 << 16)
+/* Apply rule to list */
+#define XATTR_MAP_FLAG_SERVER  (1 << 17)
+/* Apply rule to all */
+#define XATTR_MAP_FLAG_ALL   (XATTR_MAP_FLAG_SERVER | XATTR_MAP_FLAG_CLIENT)
+
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2048

Quoted 2 lines
tools/virtiofsd/passthrough_ll.c +2 -0
+/* Last rule in the XATTR_MAP */
+#define XATTR_MAP_FLAG_LAST    (1 << 30)

I see that you are using bit positions for flags. Not clear why you used bit 0,1,2 and then jumped to 16,17 and then to 30. May be you are doing some sort of reservation of bits. Will be nice to explain that a bit so that next person modifying it can use bits from correct pool.

I've added a 'types' and a 'scopes' comment pair to hopefully make it clear how I split it up.

Quoted 13 lines
Quoted 13 lines
tools/virtiofsd/passthrough_ll.c +13 -0
+
+static XattrMapEntry *add_xattrmap_entry(XattrMapEntry *orig_map,
+                                         size_t *nentries,
+                                         const XattrMapEntry *new_entry)
+{
+    XattrMapEntry *res = g_realloc_n(orig_map, ++*nentries,
+                                     sizeof(XattrMapEntry));
+    res[*nentries - 1] = *new_entry;
+
+    return res;
+}
+
+static void free_xattrmap(XattrMapEntry *map)
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2067

Quoted 6 lines
tools/virtiofsd/passthrough_ll.c +6 -0
+{
+    XattrMapEntry *curr = map;
+
+    if (!map) {
+        return;
+    };

; after } is not needed.

Gone.

Quoted 10 lines
Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+
+    do {
+        g_free(curr->key);
+        g_free(curr->prepend);
+    } while (!(curr++->flags & XATTR_MAP_FLAG_LAST));
+
+    g_free(map);
+}
+
+static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2081

Quoted 4 lines
tools/virtiofsd/passthrough_ll.c +4 -0
+{
+    XattrMapEntry *res = NULL;
+    XattrMapEntry tmp_entry;
+    size_t nentries = 0;

If you are calculating number of entries (nentries), may be this could be stored in lo_data so that can be later used to free entries or loop through rules etc.

Done; and that removes the need for _LAST.

Quoted 9 lines
Quoted 9 lines
tools/virtiofsd/passthrough_ll.c +9 -0
+    const char *map = lo->xattrmap;
+    const char *tmp;
+
+    while (*map) {
+        char sep;
+
+        if (isspace(*map)) {
+            map++;
+            continue;
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2096

Quoted 6 lines
tools/virtiofsd/passthrough_ll.c +6 -0
+        }
+        /* The separator is the first non-space of the rule */
+        sep = *map++;
+        if (!sep) {
+            break;
+        }

When can sep be NULL? In that case while loop will not even continue.

The end of the rule list.

Quoted 88 lines
Quoted 75 lines
tools/virtiofsd/passthrough_ll.c +75 -0
+
+        /* Start of 'type' */
+        if (strstart(map, "prefix", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_PREFIX;
+        } else if (strstart(map, "ok", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_OK;
+        } else if (strstart(map, "bad", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_END_BAD;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected type;"
+                     "Expecting 'prefix', 'ok', or 'bad' in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of type field of rule %zu\n",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+
+        /* Start of 'scope' */
+        if (strstart(map, "client", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_CLIENT;
+        } else if (strstart(map, "server", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_SERVER;
+        } else if (strstart(map, "all", &map)) {
+            tmp_entry.flags |= XATTR_MAP_FLAG_ALL;
+        } else {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Unexpected scope;"
+                     " Expecting 'client', 'server', or 'all', in rule %zu\n",
+                     __func__, nentries);
+            exit(1);
+        }
+
+        if (*map++ != sep) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Expecting '%c' found '%c'"
+                     " after scope in rule %zu\n",
+                     __func__, sep, *map, nentries);
+            exit(1);
+        }
+
+        /* At start of 'key' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of key field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.key = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        /* At start of 'prepend' field */
+        tmp = strchr(map, sep);
+        if (!tmp) {
+            fuse_log(FUSE_LOG_ERR,
+                     "%s: Missing '%c' at end of prepend field of rule %zu",
+                     __func__, sep, nentries);
+            exit(1);
+        }
+        tmp_entry.prepend = g_strndup(map, tmp - map);
+        map = tmp + 1;
+
+        lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                                &tmp_entry);
+        /* End of rule - go around again for another rule */
+    }
+
+    if (!nentries) {
+        fuse_log(FUSE_LOG_ERR, "Empty xattr map\n");

tools/virtiofsd/passthrough_ll.c static void lo_flock(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, · line 2181

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+        exit(1);
+    }
+
+    /* Add a terminator to error in cases the user hasn't specified */
+    tmp_entry.flags = XATTR_MAP_FLAG_ALL | XATTR_MAP_FLAG_END_BAD |
+                      XATTR_MAP_FLAG_LAST;
+    tmp_entry.key = g_strdup("");
+    tmp_entry.prepend = g_strdup("");
+    lo->xattr_map_list = add_xattrmap_entry(lo->xattr_map_list, &nentries,
+                                            &tmp_entry);

Not sure why this default rule is needed when user has not specified one.

Quoted 3 lines

This seems to be equivalent of ":bad:all:::". Will this not block all the xattrs which have not been caught by previous rules. And user probably did not want it.

I might be able to get rid of that now; my preference is to tell the users they should be explicit about what happens.

Dave

Quoted 3 lines

Thanks
Vivek

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#7 lore raw

PATCH v3 2/5 tools/virtiofsd: xattr name mappings: Map client xattr names

In reply to Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names originating at the client; from get/set/remove xattr.

---
 tools/virtiofsd/passthrough_ll.c | 101 ++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 3 deletions(-)
tools/virtiofsd/passthrough_ll.c +98 -3
index f5a33014f9..57ebe17ed6 100644
@@ -2183,20 +2183,80 @@ static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
     return res;
 }
 
-static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
+/*
+ * For use with getxattr/setxattr/removexattr, where the client
+ * gives us a name and we may need to choose a different one.
+ * Allocates a buffer for the result placing it in *out_name.
+ *   If there's no change then *out_name is not set.
+ * Returns 0 on success
+ * Can return -EPERM to indicate we block a given attribute
+ *   (in which case out_name is not allocated)
+ * Can return -ENOMEM to indicate out_name couldn't be allocated.
+ */
+static int xattr_map_client(const struct lo_data *lo, const char *client_name,
+                            char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {
+        if ((cur_entry->flags & XATTR_MAP_FLAG_CLIENT) &&
+            (strstart(client_name, cur_entry->key, NULL))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -EPERM;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
+                /* Unmodified name */
+                return 0;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {
+                *out_name = g_try_malloc(strlen(client_name) +
+                                         strlen(cur_entry->prepend) + 1);
+                if (!*out_name) {
+                    return -ENOMEM;
+                }
+                sprintf(*out_name, "%s%s", cur_entry->prepend, client_name);
+                return 0;
+            }
+        }
+    }
+
+    /* Shouldn't get here - rules should have an END_* */
+    abort();
+}
+
+static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *in_name,
                         size_t size)
 {
     struct lo_data *lo = lo_data(req);
     char *value = NULL;
     char procname[64];
+    const char *name;
+    char *mapped_name;
     struct lo_inode *inode;
     ssize_t ret;
     int saverr;
     int fd = -1;
 
+    mapped_name = NULL;
+    name = in_name;
+    if (lo->xattrmap) {
+        ret = xattr_map_client(lo, in_name, &mapped_name);
+        if (ret < 0) {
+            if (ret == -EPERM) {
+                ret = -ENODATA;
+            }
+            fuse_reply_err(req, -ret);
+            return;
+        }
+        if (mapped_name) {
+            name = mapped_name;
+        }
+    }
+
     inode = lo_inode(req, ino);
     if (!inode) {
         fuse_reply_err(req, EBADF);
+        g_free(mapped_name);
         return;
     }
 
@@ -2261,6 +2321,7 @@ out_err:
     saverr = errno;
 out:
     fuse_reply_err(req, saverr);
+    g_free(mapped_name);
     goto out_free;
 }
 
@@ -2338,19 +2399,35 @@ out:
     goto out_free;
 }
 
-static void lo_setxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
+static void lo_setxattr(fuse_req_t req, fuse_ino_t ino, const char *in_name,
                         const char *value, size_t size, int flags)
 {
     char procname[64];
+    const char *name;
+    char *mapped_name;
     struct lo_data *lo = lo_data(req);
     struct lo_inode *inode;
     ssize_t ret;
     int saverr;
     int fd = -1;
 
+    mapped_name = NULL;
+    name = in_name;
+    if (lo->xattrmap) {
+        ret = xattr_map_client(lo, in_name, &mapped_name);
+        if (ret < 0) {
+            fuse_reply_err(req, -ret);
+            return;
+        }
+        if (mapped_name) {
+            name = mapped_name;
+        }
+    }
+
     inode = lo_inode(req, ino);
     if (!inode) {
         fuse_reply_err(req, EBADF);
+        g_free(mapped_name);
         return;
     }
 
@@ -2385,21 +2462,38 @@ out:
     }
 
     lo_inode_put(lo, &inode);
+    g_free(mapped_name);
     fuse_reply_err(req, saverr);
 }
 
-static void lo_removexattr(fuse_req_t req, fuse_ino_t ino, const char *name)
+static void lo_removexattr(fuse_req_t req, fuse_ino_t ino, const char *in_name)
 {
     char procname[64];
+    const char *name;
+    char *mapped_name;
     struct lo_data *lo = lo_data(req);
     struct lo_inode *inode;
     ssize_t ret;
     int saverr;
     int fd = -1;
 
+    mapped_name = NULL;
+    name = in_name;
+    if (lo->xattrmap) {
+        ret = xattr_map_client(lo, in_name, &mapped_name);
+        if (ret < 0) {
+            fuse_reply_err(req, -ret);
+            return;
+        }
+        if (mapped_name) {
+            name = mapped_name;
+        }
+    }
+
     inode = lo_inode(req, ino);
     if (!inode) {
         fuse_reply_err(req, EBADF);
+        g_free(mapped_name);
         return;
     }
 
@@ -2434,6 +2528,7 @@ out:
     }
 
     lo_inode_put(lo, &inode);
+    g_free(mapped_name);
     fuse_reply_err(req, saverr);
 }
 
Signature · 1 line
2.28.0
Reviewed-by on v3 2/5
#8 lore raw

Re: [PATCH v3 2/5] tools/virtiofsd: xattr name mappings: Map client xattr names

Quoted 8 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names originating at the client; from get/set/remove xattr.

---
 tools/virtiofsd/passthrough_ll.c | 101 ++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 3 deletions(-)
#9 lore raw

Re: [PATCH v3 2/5] tools/virtiofsd: xattr name mappings: Map client xattr names

In reply to Dr. David Alan Gilbert (git)
Quoted 34 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names originating at the client; from get/set/remove xattr.

---
 tools/virtiofsd/passthrough_ll.c | 101 ++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 3 deletions(-)
tools/virtiofsd/passthrough_ll.c +16 -1
index f5a33014f9..57ebe17ed6 100644
@@ -2183,20 +2183,80 @@ static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
     return res;
 }
 
-static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
+/*
+ * For use with getxattr/setxattr/removexattr, where the client
+ * gives us a name and we may need to choose a different one.
+ * Allocates a buffer for the result placing it in *out_name.
+ *   If there's no change then *out_name is not set.
+ * Returns 0 on success
+ * Can return -EPERM to indicate we block a given attribute
+ *   (in which case out_name is not allocated)
+ * Can return -ENOMEM to indicate out_name couldn't be allocated.
+ */
+static int xattr_map_client(const struct lo_data *lo, const char *client_name,
+                            char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2211

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +10 -0
+        if ((cur_entry->flags & XATTR_MAP_FLAG_CLIENT) &&
+            (strstart(client_name, cur_entry->key, NULL))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -EPERM;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
+                /* Unmodified name */
+                return 0;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {

I am wondering why do have "END" substring in BAD and OK flags while we don't have one in PREFIX flag. IOW, why not simply call these flags as XATTR_MAP_FLAG_OK and XATTR_MAP_FLAG_BAD respectively.

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2213

Quoted 2 lines
tools/virtiofsd/passthrough_ll.c +2 -0
+                *out_name = g_try_malloc(strlen(client_name) +
+                                         strlen(cur_entry->prepend) + 1);

Should we check for cur_entry->prepend to be NULL before we try to allocate out_name. One could say.

"prefix:client:trusted.::". In that case we are not supposed to prefix anything?

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2225

Quoted 12 lines
tools/virtiofsd/passthrough_ll.c +12 -0
+                if (!*out_name) {
+                    return -ENOMEM;
+                }
+                sprintf(*out_name, "%s%s", cur_entry->prepend, client_name);
+                return 0;
+            }
+        }
+    }
+
+    /* Shouldn't get here - rules should have an END_* */
+    abort();
+}

Thanks
Vivek

#10 lore raw

Re: [PATCH v3 2/5] tools/virtiofsd: xattr name mappings: Map client xattr names

Quoted 41 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 40 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names originating at the client; from get/set/remove xattr.

---
 tools/virtiofsd/passthrough_ll.c | 101 ++++++++++++++++++++++++++++++-
 1 file changed, 98 insertions(+), 3 deletions(-)
tools/virtiofsd/passthrough_ll.c +22 -1
index f5a33014f9..57ebe17ed6 100644
@@ -2183,20 +2183,80 @@ static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
     return res;
 }
 
-static void lo_getxattr(fuse_req_t req, fuse_ino_t ino, const char *name,
+/*
+ * For use with getxattr/setxattr/removexattr, where the client
+ * gives us a name and we may need to choose a different one.
+ * Allocates a buffer for the result placing it in *out_name.
+ *   If there's no change then *out_name is not set.
+ * Returns 0 on success
+ * Can return -EPERM to indicate we block a given attribute
+ *   (in which case out_name is not allocated)
+ * Can return -ENOMEM to indicate out_name couldn't be allocated.
+ */
+static int xattr_map_client(const struct lo_data *lo, const char *client_name,
+                            char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {
+        if ((cur_entry->flags & XATTR_MAP_FLAG_CLIENT) &&
+            (strstart(client_name, cur_entry->key, NULL))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -EPERM;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2211

Quoted 4 lines
tools/virtiofsd/passthrough_ll.c +4 -0
+                /* Unmodified name */
+                return 0;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {

I am wondering why do have "END" substring in BAD and OK flags while we don't have one in PREFIX flag. IOW, why not simply call these flags as XATTR_MAP_FLAG_OK and XATTR_MAP_FLAG_BAD respectively.

OK, the END_'s have gone.

Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2213

Quoted 2 lines
tools/virtiofsd/passthrough_ll.c +2 -0
+                *out_name = g_try_malloc(strlen(client_name) +
+                                         strlen(cur_entry->prepend) + 1);

Should we check for cur_entry->prepend to be NULL before we try to allocate out_name. One could say.

"prefix:client:trusted.::". In that case we are not supposed to prefix anything?

We shouldn't need to do the NULL check; cur_entry->prepend should = "" in that case, not NULL.

Dave

Quoted 15 lines

tools/virtiofsd/passthrough_ll.c static XattrMapEntry *parse_xattrmap(struct lo_data *lo) · line 2225

Quoted 12 lines
tools/virtiofsd/passthrough_ll.c +12 -0
+                if (!*out_name) {
+                    return -ENOMEM;
+                }
+                sprintf(*out_name, "%s%s", cur_entry->prepend, client_name);
+                return 0;
+            }
+        }
+    }
+
+    /* Shouldn't get here - rules should have an END_* */
+    abort();
+}

Thanks
Vivek

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#11 lore raw

PATCH v3 3/5 tools/virtiofsd: xattr name mappings: Map server xattr names

In reply to Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names coming from the server, i.e. the host filesystem; currently this is only from listxattr.

---
 tools/virtiofsd/passthrough_ll.c | 89 ++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
tools/virtiofsd/passthrough_ll.c +89 -0
index 57ebe17ed6..8406a2ae86 100644
@@ -2220,6 +2220,43 @@ static int xattr_map_client(const struct lo_data *lo, const char *client_name,
         }
     }
 
+    /* Shouldn't get here - rules should have an END_* - check parse_xattrmap */
+    abort();
+}
+
+/*
+ * For use with listxattr where the server fs gives us a name and we may need
+ * to sanitize this for the client.
+ * Returns a pointer to the result in *out_name
+ *   This is always the original string or the current string with some prefix
+ *   removed; no reallocation is done.
+ * Returns 0 on success
+ * Can return -ENODATA to indicate the name should be dropped from the list.
+ */
+static int xattr_map_server(const struct lo_data *lo, const char *server_name,
+                            const char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+    const char *end;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {
+        if ((cur_entry->flags & XATTR_MAP_FLAG_SERVER) &&
+            (strstart(server_name, cur_entry->prepend, &end))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -ENODATA;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
+                *out_name = server_name;
+                return 0;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {
+                /* Remove prefix */
+                *out_name = end;
+                return 0;
+            }
+        }
+    }
+
     /* Shouldn't get here - rules should have an END_* */
     abort();
 }
@@ -2378,8 +2415,60 @@ static void lo_listxattr(fuse_req_t req, fuse_ino_t ino, size_t size)
         if (ret == 0) {
             goto out;
         }
+
+        if (lo->xattr_map_list) {
+            /*
+             * Map the names back, some attributes might be dropped,
+             * some shortened, but not increased, so we shouldn't
+             * run out of room.
+             */
+            size_t out_index, in_index;
+            out_index = 0;
+            in_index = 0;
+            while (in_index < ret) {
+                const char *map_out;
+                char *in_ptr = value + in_index;
+                /* Length of current attribute name */
+                size_t in_len = strlen(value + in_index) + 1;
+
+                int mapret = xattr_map_server(lo, in_ptr, &map_out);
+                if (mapret != -ENODATA && mapret != 0) {
+                    /* Shouldn't happen */
+                    saverr = -mapret;
+                    goto out;
+                }
+                if (mapret == 0) {
+                    /* Either unchanged, or truncated */
+                    size_t out_len;
+                    if (map_out != in_ptr) {
+                        /* +1 copies the NIL */
+                        out_len = strlen(map_out) + 1;
+                    } else {
+                        /* No change */
+                        out_len = in_len;
+                    }
+                    /*
+                     * Move result along, may still be needed for an unchanged
+                     * entry if a previous entry was changed.
+                     */
+                    memmove(value + out_index, map_out, out_len);
+
+                    out_index += out_len;
+                }
+                in_index += in_len;
+            }
+            ret = out_index;
+            if (ret == 0) {
+                goto out;
+            }
+        }
         fuse_reply_buf(req, value, ret);
     } else {
+        /*
+         * xattrmap only ever shortens the result,
+         * so we don't need to do anything clever with the
+         * allocation length here.
+         */
         fuse_reply_xattr(req, ret);
     }
 out_free:
Signature · 1 line
2.28.0
#12 lore raw

[Virtio-fs] Puzzle about rootflags, restorecon "operation not supported"

The 'mainline generic' version of the latest kernel fails to parse 'fuse' virtiofs options on root kernel boots.  Not only dax, but all the 'fuse' usual ones as well.  (None of default_permissions,allow_other,user_id=0 etc are accepted).

Is that what you expected of the mainline kernel?  I learned of this when trying to understand this further problem:

SELinux's 'restorecon' and other attempts to change security attributes then fails with "operation not supported".

I could only get sshd able to accept connections by changing SELinux to permissive.  The commands necessary to change the security attributes failed with the 'operation not supported' issue.

Ideas?

#13 lore raw

Re: [Virtio-fs] Puzzle about rootflags, restorecon "operation not supported"

Quoted 8 lines · Harry G. Coin

The 'mainline generic' version of the latest kernel fails to parse 'fuse' virtiofs options on root kernel boots.  Not only dax, but all the 'fuse' usual ones as well.  (None of default_permissions,allow_other,user_id=0 etc are accepted).

Is that what you expected of the mainline kernel?  I learned of this when trying to understand this further problem:

Quoted 8 lines

SELinux's 'restorecon' and other attempts to change security attributes then fails with "operation not supported".

I could only get sshd able to accept connections by changing SELinux to permissive.  The commands necessary to change the security attributes failed with the 'operation not supported' issue.

Ideas?

Hi Harry,
FUSE mount options were removed from virtiofs at one point. I think this is expected. The reason is because the options you listed are the default for virtiofs and don't need to be specified explicitly.

I have CCed Miklos in case things have changed or I missed something.

Stefan

#14 lore raw

Re: [Virtio-fs] Puzzle about rootflags, restorecon "operation not supported"

Quoted 17 lines · Stefan Hajnoczi
Quoted 15 lines · Harry G. Coin

The 'mainline generic' version of the latest kernel fails to parse 'fuse' virtiofs options on root kernel boots. Not only dax, but all the 'fuse' usual ones as well. (None of default_permissions,allow_other,user_id=0 etc are accepted).

Is that what you expected of the mainline kernel? I learned of this when trying to understand this further problem:

SELinux's 'restorecon' and other attempts to change security attributes then fails with "operation not supported".

I could only get sshd able to accept connections by changing SELinux to permissive. The commands necessary to change the security attributes failed with the 'operation not supported' issue.

Quoted 8 lines
Quoted 1 line

Ideas?

Hi Harry,
FUSE mount options were removed from virtiofs at one point. I think this is expected. The reason is because the options you listed are the default for virtiofs and don't need to be specified explicitly.

I have CCed Miklos in case things have changed or I missed something.

Removing the fuse mount options is expected. Removing the selinux option is probably a bug.

With the DAX patches the option parsing was restored. Was the selinux option parsing also restored?

Thanks,
Miklos

Quoted 2 lines

Stefan

#15 lore raw

Re: [Virtio-fs] Puzzle about rootflags, restorecon "operation not supported"

Quoted 23 lines · Miklos Szeredi
Quoted 22 lines · Stefan Hajnoczi
Quoted 8 lines · Harry G. Coin

The 'mainline generic' version of the latest kernel fails to parse 'fuse' virtiofs options on root kernel boots. Not only dax, but all the 'fuse' usual ones as well. (None of default_permissions,allow_other,user_id=0 etc are accepted).

Is that what you expected of the mainline kernel? I learned of this when trying to understand this further problem:

Quoted 8 lines

SELinux's 'restorecon' and other attempts to change security attributes then fails with "operation not supported".

I could only get sshd able to accept connections by changing SELinux to permissive. The commands necessary to change the security attributes failed with the 'operation not supported' issue.

Ideas?

Hi Harry,
FUSE mount options were removed from virtiofs at one point. I think this is expected. The reason is because the options you listed are the default for virtiofs and don't need to be specified explicitly.

Quoted 6 lines
Quoted 1 line

I have CCed Miklos in case things have changed or I missed something.

Removing the fuse mount options is expected. Removing the selinux option is probably a bug.

With the DAX patches the option parsing was restored. Was the selinux option parsing also restored?

I suggest a small effort at these user-facing doc updates would help adoption.  A lot.

What would help the most on the gitlab web page is: a roster of all guest side mount options, with notes as to deprecated and no longer implemented options, and notes if an option requires a particular host-side xml or qemu or virtiofsd setting, and notes if guest kernel option is not available at kernel boot time.  If there's an 'selinux option' somewhere to be found I at least missed it on the official virtiofs doc page.

The kernel.org 'filesystems latest' virtiofs doc is dated 2019. It contains "virtiofs supports general VFS mount options, for example, remount, ro, rw, context, etc. It also supports FUSE mount options." Which appears to be out of date (file systems using those options fail to mount/boot).

The libvirt page discusses host setup xml, but doesn't include much of the option suite virtiofsd's manpage offers.

Both kernel.org and libvirt reference the gitlab page, nowhere in which do I find references to fedora past v 29 (now up to 33...).

I suggest the 'standalone virtio-fs' page include a link to two different known-good kernel '.config' files.  One  that when compiled will deliver the virtio-fs features built into the kernel that supports booting without an initrd, and another that enables as modules the necessary for initrd or non-virtiofsd-as-root use.  Earlier I offered the updates to dracut necessary to create an initrd using the kernel command grammar other packages use, but those should by default support SELinux , so the default flags need updating.

Thanks for the focus and your work on this!

Harry Coin

Quoted 6 lines

Thanks,
Miklos

Quoted 1 line

Stefan

Reviewed-by on v3 3/5
#16 lore raw

Re: [PATCH v3 3/5] tools/virtiofsd: xattr name mappings: Map server xattr names

In reply to Dr. David Alan Gilbert (git)
Quoted 9 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names coming from the server, i.e. the host filesystem; currently this is only from listxattr.

---
 tools/virtiofsd/passthrough_ll.c | 89 ++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
#17 lore raw

Re: [PATCH v3 3/5] tools/virtiofsd: xattr name mappings: Map server xattr names

In reply to Dr. David Alan Gilbert (git)
Quoted 47 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names coming from the server, i.e. the host filesystem; currently this is only from listxattr.

---
 tools/virtiofsd/passthrough_ll.c | 89 ++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
tools/virtiofsd/passthrough_ll.c +29 -0
index 57ebe17ed6..8406a2ae86 100644
@@ -2220,6 +2220,43 @@ static int xattr_map_client(const struct lo_data *lo, const char *client_name,
         }
     }
 
+    /* Shouldn't get here - rules should have an END_* - check parse_xattrmap */
+    abort();
+}
+
+/*
+ * For use with listxattr where the server fs gives us a name and we may need
+ * to sanitize this for the client.
+ * Returns a pointer to the result in *out_name
+ *   This is always the original string or the current string with some prefix
+ *   removed; no reallocation is done.
+ * Returns 0 on success
+ * Can return -ENODATA to indicate the name should be dropped from the list.
+ */
+static int xattr_map_server(const struct lo_data *lo, const char *server_name,
+                            const char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+    const char *end;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {
+        if ((cur_entry->flags & XATTR_MAP_FLAG_SERVER) &&
+            (strstart(server_name, cur_entry->prepend, &end))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -ENODATA;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
+                *out_name = server_name;
+                return 0;
+            }

tools/virtiofsd/passthrough_ll.c static int xattr_map_client(const struct lo_data *lo, const char *client_name, · line 2261

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +8 -0
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {
+                /* Remove prefix */
+                *out_name = end;
+                return 0;
+            }
+        }
+    }
+
     /* Shouldn't get here - rules should have an END_* */
     abort();

I am wondering why to put that restriction. If none of the rules match, can't we just return as nothing has to be done.

Quoted 50 lines
tools/virtiofsd/passthrough_ll.c +45 -0
 }
@@ -2378,8 +2415,60 @@ static void lo_listxattr(fuse_req_t req, fuse_ino_t ino, size_t size)
         if (ret == 0) {
             goto out;
         }
+
+        if (lo->xattr_map_list) {
+            /*
+             * Map the names back, some attributes might be dropped,
+             * some shortened, but not increased, so we shouldn't
+             * run out of room.
+             */
+            size_t out_index, in_index;
+            out_index = 0;
+            in_index = 0;
+            while (in_index < ret) {
+                const char *map_out;
+                char *in_ptr = value + in_index;
+                /* Length of current attribute name */
+                size_t in_len = strlen(value + in_index) + 1;
+
+                int mapret = xattr_map_server(lo, in_ptr, &map_out);
+                if (mapret != -ENODATA && mapret != 0) {
+                    /* Shouldn't happen */
+                    saverr = -mapret;
+                    goto out;
+                }
+                if (mapret == 0) {
+                    /* Either unchanged, or truncated */
+                    size_t out_len;
+                    if (map_out != in_ptr) {
+                        /* +1 copies the NIL */
+                        out_len = strlen(map_out) + 1;
+                    } else {
+                        /* No change */
+                        out_len = in_len;
+                    }
+                    /*
+                     * Move result along, may still be needed for an unchanged
+                     * entry if a previous entry was changed.
+                     */
+                    memmove(value + out_index, map_out, out_len);
+
+                    out_index += out_len;
+                }
+                in_index += in_len;
+            }
+            ret = out_index;
+            if (ret == 0) {
+                goto out;

tools/virtiofsd/passthrough_ll.c static void lo_listxattr(fuse_req_t req, fuse_ino_t ino, size_t size) · line 2472

Quoted 10 lines
tools/virtiofsd/passthrough_ll.c +7 -0
+            }
+        }
         fuse_reply_buf(req, value, ret);
     } else {
+        /*
+         * xattrmap only ever shortens the result,
+         * so we don't need to do anything clever with the
+         * allocation length here.
+         */
         fuse_reply_xattr(req, ret);

Hmmm.., so this code returns the length of buffer which will fit xattrs. So we will will changing the semantics a bit. Instead of returning the exact size of buffer needed, we will be returning max size. I hope its not a problem. Fixing it will be too expensive I guess.

Thanks
Vivek

#18 lore raw

Re: [PATCH v3 3/5] tools/virtiofsd: xattr name mappings: Map server xattr names

Quoted 53 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 52 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Map xattr names coming from the server, i.e. the host filesystem; currently this is only from listxattr.

---
 tools/virtiofsd/passthrough_ll.c | 89 ++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
tools/virtiofsd/passthrough_ll.c +34 -0
index 57ebe17ed6..8406a2ae86 100644
@@ -2220,6 +2220,43 @@ static int xattr_map_client(const struct lo_data *lo, const char *client_name,
         }
     }
 
+    /* Shouldn't get here - rules should have an END_* - check parse_xattrmap */
+    abort();
+}
+
+/*
+ * For use with listxattr where the server fs gives us a name and we may need
+ * to sanitize this for the client.
+ * Returns a pointer to the result in *out_name
+ *   This is always the original string or the current string with some prefix
+ *   removed; no reallocation is done.
+ * Returns 0 on success
+ * Can return -ENODATA to indicate the name should be dropped from the list.
+ */
+static int xattr_map_server(const struct lo_data *lo, const char *server_name,
+                            const char **out_name)
+{
+    const XattrMapEntry *cur_entry;
+    const char *end;
+
+    for (cur_entry = lo->xattr_map_list; ; cur_entry++) {
+        if ((cur_entry->flags & XATTR_MAP_FLAG_SERVER) &&
+            (strstart(server_name, cur_entry->prepend, &end))) {
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_BAD) {
+                return -ENODATA;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_END_OK) {
+                *out_name = server_name;
+                return 0;
+            }
+            if (cur_entry->flags & XATTR_MAP_FLAG_PREFIX) {
+                /* Remove prefix */
+                *out_name = end;
+                return 0;
+            }
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static int xattr_map_client(const struct lo_data *lo, const char *client_name, · line 2261

Quoted 5 lines
tools/virtiofsd/passthrough_ll.c +3 -0
+        }
+    }
+
     /* Shouldn't get here - rules should have an END_* */
     abort();

I am wondering why to put that restriction. If none of the rules match, can't we just return as nothing has to be done.

I always add a terminator in the parse as either a bad/ok, and was just enforcing it - but I've changed it to a return -ENODATA that's probably safer than the abort().

Quoted 57 lines
Quoted 57 lines
tools/virtiofsd/passthrough_ll.c +50 -0
 }
@@ -2378,8 +2415,60 @@ static void lo_listxattr(fuse_req_t req, fuse_ino_t ino, size_t size)
         if (ret == 0) {
             goto out;
         }
+
+        if (lo->xattr_map_list) {
+            /*
+             * Map the names back, some attributes might be dropped,
+             * some shortened, but not increased, so we shouldn't
+             * run out of room.
+             */
+            size_t out_index, in_index;
+            out_index = 0;
+            in_index = 0;
+            while (in_index < ret) {
+                const char *map_out;
+                char *in_ptr = value + in_index;
+                /* Length of current attribute name */
+                size_t in_len = strlen(value + in_index) + 1;
+
+                int mapret = xattr_map_server(lo, in_ptr, &map_out);
+                if (mapret != -ENODATA && mapret != 0) {
+                    /* Shouldn't happen */
+                    saverr = -mapret;
+                    goto out;
+                }
+                if (mapret == 0) {
+                    /* Either unchanged, or truncated */
+                    size_t out_len;
+                    if (map_out != in_ptr) {
+                        /* +1 copies the NIL */
+                        out_len = strlen(map_out) + 1;
+                    } else {
+                        /* No change */
+                        out_len = in_len;
+                    }
+                    /*
+                     * Move result along, may still be needed for an unchanged
+                     * entry if a previous entry was changed.
+                     */
+                    memmove(value + out_index, map_out, out_len);
+
+                    out_index += out_len;
+                }
+                in_index += in_len;
+            }
+            ret = out_index;
+            if (ret == 0) {
+                goto out;
+            }
+        }
         fuse_reply_buf(req, value, ret);
     } else {
+        /*
+         * xattrmap only ever shortens the result,
+         * so we don't need to do anything clever with the
Quoted 8 lines

tools/virtiofsd/passthrough_ll.c static void lo_listxattr(fuse_req_t req, fuse_ino_t ino, size_t size) · line 2472

Quoted 3 lines
tools/virtiofsd/passthrough_ll.c +2 -0
+         * allocation length here.
+         */
         fuse_reply_xattr(req, ret);

Hmmm.., so this code returns the length of buffer which will fit xattrs. So we will will changing the semantics a bit. Instead of returning the exact size of buffer needed, we will be returning max size. I hope its not a problem. Fixing it will be too expensive I guess.

Right, although the semantics are fuzzy anyway since someone can come along and add/remove an xattr between the listxattr calls.

Dave

Quoted 2 lines

Thanks
Vivek

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#19 lore raw

PATCH v3 4/5 tools/virtiofsd: xattr name mapping examples

In reply to Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
docs/tools/virtiofsd.rst +50 -0
index a3a120da2f..5cb64612ed 100644
@@ -163,6 +163,56 @@ in which case a 'server' rule will always match on all names from
 the server.
 
 
+xattr-mapping Examples
+----------------------
+
+1) Prefix all attributes with 'user.virtiofs.'
+
+::
+
+-o xattrmap=":prefix:all::user.virtiofs.::bad:all:::"
+
+
+This uses two rules, using : as the field separator;
+the first rule prefixes and strips 'user.virtiofs.',
+the second rule hides any non-prefixed attributes that
+the host set.
+
+2) Prefix 'trusted.' attributes, allow others through
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.
+The second rule hides unprefixed 'trusted.' attributes
+on the host.
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.
+
+3) Hide 'security.' attributes, and allow everything else
+
+::
+
+    "/bad/all/security./security./
+     /ok/all///'
+
+The first rule combines what could be separate client and server
+rules into a single 'all' rule, matching 'security.' in either
+client arguments or lists returned from the host.  This stops
+the client seeing any 'security.' attributes on the server and
+stops it setting any.
+
 Examples
 --------
 
Signature · 1 line
2.28.0
Reviewed-by on v3 4/5
#20 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

Quoted 8 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
#21 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

In reply to Dr. David Alan Gilbert (git)
Quoted 16 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
docs/tools/virtiofsd.rst +0 -0
index a3a120da2f..5cb64612ed 100644
@@ -163,6 +163,56 @@ in which case a 'server' rule will always match on all names from
 the server.
 

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 174

Quoted 10 lines
docs/tools/virtiofsd.rst +9 -0
 
+xattr-mapping Examples
+----------------------
+
+1) Prefix all attributes with 'user.virtiofs.'
+
+::
+
+-o xattrmap=":prefix:all::user.virtiofs.::bad:all:::"
+

Staring at rule.

":bad:all:::"

and trying to map this to ":type:scope:key:prepend:"

type==bad
scope==all
key==""
prepend==""

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 179

Quoted 5 lines
docs/tools/virtiofsd.rst +5 -0
+
+This uses two rules, using : as the field separator;
+the first rule prefixes and strips 'user.virtiofs.',
+the second rule hides any non-prefixed attributes that
+the host set.

What is non-prefixed xattr in this context. If client sends "security.selinux", is prefixed or non-prefixed.

Documentation in first patch said.

+- 'bad' - If a client tries to use this name it's
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.

I am not sure which xattr name will be blocked if key=="" and prepend==""

Quoted 6 lines
docs/tools/virtiofsd.rst +6 -0
+
+2) Prefix 'trusted.' attributes, allow others through
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 10 lines
docs/tools/virtiofsd.rst +10 -0
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 2 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Thanks
Vivek

Quoted 20 lines
docs/tools/virtiofsd.rst +14 -0
+
+3) Hide 'security.' attributes, and allow everything else
+
+::
+
+    "/bad/all/security./security./
+     /ok/all///'
+
+The first rule combines what could be separate client and server
+rules into a single 'all' rule, matching 'security.' in either
+client arguments or lists returned from the host.  This stops
+the client seeing any 'security.' attributes on the server and
+stops it setting any.
+
 Examples
 --------
 
Signature · 1 line
2.28.0
#22 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

Quoted 30 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 16 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
docs/tools/virtiofsd.rst +0 -0
index a3a120da2f..5cb64612ed 100644
@@ -163,6 +163,56 @@ in which case a 'server' rule will always match on all names from
 the server.
 

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 174

Quoted 10 lines
docs/tools/virtiofsd.rst +9 -0
 
+xattr-mapping Examples
+----------------------
+
+1) Prefix all attributes with 'user.virtiofs.'
+
+::
+
+-o xattrmap=":prefix:all::user.virtiofs.::bad:all:::"
+

Staring at rule.

Quoted 8 lines

":bad:all:::"

and trying to map this to ":type:scope:key:prepend:"

type==bad
scope==all
key==""
prepend==""

Correct.

Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 179

Quoted 5 lines
docs/tools/virtiofsd.rst +5 -0
+
+This uses two rules, using : as the field separator;
+the first rule prefixes and strips 'user.virtiofs.',
+the second rule hides any non-prefixed attributes that
+the host set.

What is non-prefixed xattr in this context. If client sends "security.selinux", is prefixed or non-prefixed.

Note that anything originating at the client (i.e. starting with "") will get caught by the first rule and be prefixed with 'user.virtiofs.' This second rule will only be triggered by an xattr name coming from the server (i.e a listxattr) for a name that doesn't begin with user.virtiofs. (i.e. didn't match the 1st rule for a server xattr). This makes sure that only guest created xattr's (that were set and had a prefix added) are then visible to the guest.

Quoted 7 lines

Documentation in first patch said.

+- 'bad' - If a client tries to use this name it's

(quoted diff) +2 -0
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.

I am not sure which xattr name will be blocked if key=="" and prepend==""

All of them; they're still matched at the beginning; as the first patch says 'It maybe empty in which case a 'client' rule will always match on client names'

Quoted 11 lines
Quoted 11 lines
docs/tools/virtiofsd.rst +11 -0
+2) Prefix 'trusted.' attributes, allow others through
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

prefixed not replaced; so it'll turn "trusted." into "user.virtiofs.trusted." and strip it back off for listxattr.

Quoted 6 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 2 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

No, the first rule will only prefix strings provided by the guest and strip strings provided by the server. This rule hides existing server 'trusted.' xattrs - so if the guest sets trusted.foo it's not confused by also seeing a server trusted.foo

Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Right; and that's dangerous, because a non-privileged guest process can set a user. xattr; so a non-priv guest process could set user.virtiofs.trusted.foo and then it would get read back and be used as trusted.foo that has an impact on priviliged processes.

Dave

Quoted 23 lines

Thanks
Vivek

Quoted 20 lines
docs/tools/virtiofsd.rst +14 -0
+
+3) Hide 'security.' attributes, and allow everything else
+
+::
+
+    "/bad/all/security./security./
+     /ok/all///'
+
+The first rule combines what could be separate client and server
+rules into a single 'all' rule, matching 'security.' in either
+client arguments or lists returned from the host.  This stops
+the client seeing any 'security.' attributes on the server and
+stops it setting any.
+
 Examples
 --------
 
Signature · 1 line
2.28.0
Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#23 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

Quoted 51 lines · Dr. David Alan Gilbert
Quoted 30 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 16 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
docs/tools/virtiofsd.rst +0 -0
index a3a120da2f..5cb64612ed 100644
@@ -163,6 +163,56 @@ in which case a 'server' rule will always match on all names from
 the server.
 

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 174

Quoted 10 lines
docs/tools/virtiofsd.rst +9 -0
 
+xattr-mapping Examples
+----------------------
+
+1) Prefix all attributes with 'user.virtiofs.'
+
+::
+
+-o xattrmap=":prefix:all::user.virtiofs.::bad:all:::"
+

Staring at rule.

Quoted 8 lines

":bad:all:::"

and trying to map this to ":type:scope:key:prepend:"

type==bad
scope==all
key==""
prepend==""

Correct.

Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 179

Quoted 5 lines
docs/tools/virtiofsd.rst +5 -0
+
+This uses two rules, using : as the field separator;
+the first rule prefixes and strips 'user.virtiofs.',
+the second rule hides any non-prefixed attributes that
+the host set.

What is non-prefixed xattr in this context. If client sends "security.selinux", is prefixed or non-prefixed.

Quoted 2 lines

Note that anything originating at the client (i.e. starting with "") will get caught by the first rule and be prefixed with 'user.virtiofs.'

So first rule is.

:prefix:all::user.virtiofs.:

scope="all"
key==""
prepend="user.virtiofs."

Given scope, rule applies to both client and server. Given key is "", any income setxattr() request will be prefixed with "user.virtiofs". And how does this rule work for server (listxattr()). It will match prepend "user.virtiofs" and if that matches it will be stripped?

Quoted 5 lines

This second rule will only be triggered by an xattr name coming from the server (i.e a listxattr) for a name that doesn't begin with user.virtiofs. (i.e. didn't match the 1st rule for a server xattr). This makes sure that only guest created xattr's (that were set and had a prefix added) are then visible to the guest.

Ok, so if an xattr does not match first rule, then second rule says key="", prepend="" and that will match all xattrs. So anything which is not caught by first rule, will be caught by second rule and either rejected or filtered out.

Quoted 7 lines
Quoted 7 lines

Documentation in first patch said.

+- 'bad' - If a client tries to use this name it's

(quoted diff) +2 -0
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
Quoted 5 lines
Quoted 1 line

I am not sure which xattr name will be blocked if key=="" and prepend==""

All of them; they're still matched at the beginning; as the first patch says 'It maybe empty in which case a 'client' rule will always match on client names'

Ok.

Quoted 21 lines
Quoted 12 lines
Quoted 12 lines
docs/tools/virtiofsd.rst +11 -0
+2) Prefix 'trusted.' attributes, allow others through
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

Quoted 2 lines

prefixed not replaced; so it'll turn "trusted." into "user.virtiofs.trusted." and strip it back off for listxattr.

Ok. Got it. I am wondering how will I specify these rules so that they work in nested configuration. Say I have L0 host, L1 guest and L2 guest. Say virtiofsd0 is running on L0 and virtiofsd1 is running on L1.

I am wondering how will I specify the rules on virtiofsd0 and virtiofsd1 so that it works. Will it be same or rules are level dependent.

Quoted 21 lines
Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 3 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

No, the first rule will only prefix strings provided by the guest and strip strings provided by the server. This rule hides existing server 'trusted.' xattrs - so if the guest sets trusted.foo it's not confused by also seeing a server trusted.foo

Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Quoted 4 lines

Right; and that's dangerous, because a non-privileged guest process can set a user. xattr; so a non-priv guest process could set user.virtiofs.trusted.foo and then it would get read back and be used as trusted.foo that has an impact on priviliged processes.

Right. We don't want unpriviliged process to be able to setup user.virtiofs.trusted.*. But that's what precisely happen in a nested configuration.

In above example, L2 will set trusted.foo, virtiofsd1 wil convert it to user.virtiofs.trusted.foo and virtiofsd0 will reject it, breaking the nested virtiofs.

Thanks
Vivek

#24 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

Quoted 59 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 51 lines · Dr. David Alan Gilbert
Quoted 30 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 16 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

Add a few examples of xattrmaps to the documentation.

---
 docs/tools/virtiofsd.rst | 50 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
docs/tools/virtiofsd.rst +0 -0
index a3a120da2f..5cb64612ed 100644
@@ -163,6 +163,56 @@ in which case a 'server' rule will always match on all names from
 the server.
 

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 174

Quoted 10 lines
docs/tools/virtiofsd.rst +9 -0
 
+xattr-mapping Examples
+----------------------
+
+1) Prefix all attributes with 'user.virtiofs.'
+
+::
+
+-o xattrmap=":prefix:all::user.virtiofs.::bad:all:::"
+

Staring at rule.

Quoted 8 lines

":bad:all:::"

and trying to map this to ":type:scope:key:prepend:"

type==bad
scope==all
key==""
prepend==""

Correct.

Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 179

Quoted 5 lines
docs/tools/virtiofsd.rst +5 -0
+
+This uses two rules, using : as the field separator;
+the first rule prefixes and strips 'user.virtiofs.',
+the second rule hides any non-prefixed attributes that
+the host set.

What is non-prefixed xattr in this context. If client sends "security.selinux", is prefixed or non-prefixed.

Quoted 2 lines

Note that anything originating at the client (i.e. starting with "") will get caught by the first rule and be prefixed with 'user.virtiofs.'

So first rule is.

:prefix:all::user.virtiofs.:

Quoted 8 lines

scope="all"
key==""
prepend="user.virtiofs."

Given scope, rule applies to both client and server. Given key is "", any income setxattr() request will be prefixed with "user.virtiofs". And how does this rule work for server (listxattr()). It will match prepend "user.virtiofs" and if that matches it will be stripped?

Right; 'prefix' strips for server, adds for client (as long as you have 'all', you can select either one on it's own with server/client rather than all).

Quoted 8 lines
Quoted 7 lines

This second rule will only be triggered by an xattr name coming from the server (i.e a listxattr) for a name that doesn't begin with user.virtiofs. (i.e. didn't match the 1st rule for a server xattr). This makes sure that only guest created xattr's (that were set and had a prefix added) are then visible to the guest.

Quoted 4 lines

Ok, so if an xattr does not match first rule, then second rule says key="", prepend="" and that will match all xattrs. So anything which is not caught by first rule, will be caught by second rule and either rejected or filtered out.

Right.

Quoted 39 lines
Quoted 7 lines
Quoted 7 lines

Documentation in first patch said.

+- 'bad' - If a client tries to use this name it's

(quoted diff) +2 -0
+  denied using EPERM; when the server passes an attribute
+  name matching it's hidden.
Quoted 5 lines
Quoted 1 line

I am not sure which xattr name will be blocked if key=="" and prepend==""

All of them; they're still matched at the beginning; as the first patch says 'It maybe empty in which case a 'client' rule will always match on client names'

Ok.

Quoted 21 lines
Quoted 12 lines
Quoted 12 lines
docs/tools/virtiofsd.rst +11 -0
+2) Prefix 'trusted.' attributes, allow others through
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

Quoted 2 lines

prefixed not replaced; so it'll turn "trusted." into "user.virtiofs.trusted." and strip it back off for listxattr.

Quoted 7 lines

Ok. Got it. I am wondering how will I specify these rules so that they work in nested configuration. Say I have L0 host, L1 guest and L2 guest. Say virtiofsd0 is running on L0 and virtiofsd1 is running on L1.

I am wondering how will I specify the rules on virtiofsd0 and virtiofsd1 so that it works. Will it be same or rules are level dependent.

I'm hoping it'll be the same, see below.

Quoted 26 lines
Quoted 21 lines
Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 3 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

No, the first rule will only prefix strings provided by the guest and strip strings provided by the server. This rule hides existing server 'trusted.' xattrs - so if the guest sets trusted.foo it's not confused by also seeing a server trusted.foo

Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Quoted 4 lines

Right; and that's dangerous, because a non-privileged guest process can set a user. xattr; so a non-priv guest process could set user.virtiofs.trusted.foo and then it would get read back and be used as trusted.foo that has an impact on priviliged processes.

Quoted 7 lines

Right. We don't want unpriviliged process to be able to setup user.virtiofs.trusted.*. But that's what precisely happen in a nested configuration.

In above example, L2 will set trusted.foo, virtiofsd1 wil convert it to user.virtiofs.trusted.foo and virtiofsd0 will reject it, breaking the nested virtiofs.

So to allow nesting you need to nest the user.virtiofs. as well, not just the trusted. So either you do an all, or if you want to be more selective then I think the following would work:

 1  /prefix/client/trusted./user.virtiofs./
 2  /prefix/client/user.virtiofs./user.virtiofs./
 3  /prefix/server//user.virtiofs./
 4  /bad/server//trusted./
 5  /ok/all///

1 causes any getattr/setattr to convert 'trusted.'
to 'user.virtiofs.trusted.'
2 causes any getattr/setattr to convert 'user.virtiofs.'
to 'user.virtiofs.user.virtiofs.'
3 causes any listattr to lose a layer of user.virtiofs. 4 blocks any trusted. from the layer beneath
5 lets anything else through

(I'm trying to convince myself if we need a /bad/server//user.virtiofs.trusted. to stop the previous level being visible).

Dave

Quoted 2 lines

Thanks
Vivek

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#25 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

On Tue, Oct 20, 2020 at 08:02:37PM +0100, Dr. David Alan Gilbert wrote:

[..]

Quoted 67 lines
Quoted 23 lines
Quoted 20 lines
Quoted 11 lines
Quoted 11 lines

+2) Prefix 'trusted.' attributes, allow others through

docs/tools/virtiofsd.rst +10 -0
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

Quoted 2 lines

prefixed not replaced; so it'll turn "trusted." into "user.virtiofs.trusted." and strip it back off for listxattr.

Quoted 7 lines

Ok. Got it. I am wondering how will I specify these rules so that they work in nested configuration. Say I have L0 host, L1 guest and L2 guest. Say virtiofsd0 is running on L0 and virtiofsd1 is running on L1.

I am wondering how will I specify the rules on virtiofsd0 and virtiofsd1 so that it works. Will it be same or rules are level dependent.

I'm hoping it'll be the same, see below.

Quoted 26 lines
Quoted 21 lines
Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 3 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

No, the first rule will only prefix strings provided by the guest and strip strings provided by the server. This rule hides existing server 'trusted.' xattrs - so if the guest sets trusted.foo it's not confused by also seeing a server trusted.foo

Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Quoted 4 lines

Right; and that's dangerous, because a non-privileged guest process can set a user. xattr; so a non-priv guest process could set user.virtiofs.trusted.foo and then it would get read back and be used as trusted.foo that has an impact on priviliged processes.

Quoted 7 lines

Right. We don't want unpriviliged process to be able to setup user.virtiofs.trusted.*. But that's what precisely happen in a nested configuration.

In above example, L2 will set trusted.foo, virtiofsd1 wil convert it to user.virtiofs.trusted.foo and virtiofsd0 will reject it, breaking the nested virtiofs.

Quoted 6 lines

So to allow nesting you need to nest the user.virtiofs. as well, not just the trusted. So either you do an all, or if you want to be more selective then I think the following would work:

1 /prefix/client/trusted./user.virtiofs./
2 /prefix/client/user.virtiofs./user.virtiofs./

Ok, so basically instead of blocking user.virtiofs.trusted. from client, prefix it with "user.virtiofs." one more time. IOW, allow client to set user.virtiofs.trusted. because it will get back user.virtiofs.trusted. and not "trusted." which is ok. Now client user space can't fool client kernel with setting arbitrary user.virtiofs.trusted xattrs.

And if client kernel sends, trusted., it will get back trusted.

Only thing which can happen is that client1 sets user.virtiofs.trusted. and nested client2 will get it as trusted. So client1 user space can fool nested client's kernel. But given client1 has launched nested client2, we should be able to trust some user on client1 and make sure other users can't see this shared dir and this probably is not an issue.

Quoted 12 lines
 3  /prefix/server//user.virtiofs./
 4  /bad/server//trusted./
 5  /ok/all///

1 causes any getattr/setattr to convert 'trusted.'
to 'user.virtiofs.trusted.'
2 causes any getattr/setattr to convert 'user.virtiofs.'
to 'user.virtiofs.user.virtiofs.'
3 causes any listattr to lose a layer of user.virtiofs. 4 blocks any trusted. from the layer beneath
5 lets anything else through

Quoted 3 lines

(I'm trying to convince myself if we need a /bad/server//user.virtiofs.trusted. to stop the previous level being visible).

user.virtiofs.trusted on server will be converted to trusted., right? Can't block it otherwise L1 client breaks, isn't it?

Vivek

#26 lore raw

Re: [PATCH v3 4/5] tools/virtiofsd: xattr name mapping examples

Quoted 74 lines · * Vivek Goyal ([email protected]) wrote:

On Tue, Oct 20, 2020 at 08:02:37PM +0100, Dr. David Alan Gilbert wrote:

[..]

Quoted 71 lines
Quoted 23 lines
Quoted 20 lines
Quoted 11 lines
Quoted 11 lines

+2) Prefix 'trusted.' attributes, allow others through

docs/tools/virtiofsd.rst +10 -0
+
+::
+
+   "/prefix/all/trusted./user.virtiofs./
+    /bad/server//trusted./
+    /bad/client/user.virtiofs.//
+    /ok/all///"
+
+
+Here there are four rules, using / as the field
Quoted 8 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 195

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+separator, and also demonstrating that new lines can
+be included between rules.
+The first rule is the prefixing of 'trusted.' and
+stripping of 'user.virtiofs.'.

So this is bidrectional rule, right. For setxattr(), "trusted." will be replaced with "user.virtiofs" and for listxattr(), server will replace user.virtiofs with trusted. ?

Quoted 2 lines

prefixed not replaced; so it'll turn "trusted." into "user.virtiofs.trusted." and strip it back off for listxattr.

Quoted 7 lines

Ok. Got it. I am wondering how will I specify these rules so that they work in nested configuration. Say I have L0 host, L1 guest and L2 guest. Say virtiofsd0 is running on L0 and virtiofsd1 is running on L1.

I am wondering how will I specify the rules on virtiofsd0 and virtiofsd1 so that it works. Will it be same or rules are level dependent.

I'm hoping it'll be the same, see below.

Quoted 26 lines
Quoted 21 lines
Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 197

Quoted 3 lines
docs/tools/virtiofsd.rst +2 -0
+The second rule hides unprefixed 'trusted.' attributes
+on the host.

If host has "trusted.*", we are not hiding it and as per first rule we are converting it to "user.virtiofs.trusted.*", right? So why this second rule is needed.

No, the first rule will only prefix strings provided by the guest and strip strings provided by the server. This rule hides existing server 'trusted.' xattrs - so if the guest sets trusted.foo it's not confused by also seeing a server trusted.foo

Quoted 7 lines

docs/tools/virtiofsd.rst in which case a 'server' rule will always match on all names from · line 201

Quoted 4 lines
docs/tools/virtiofsd.rst +4 -0
+The third rule stops a guest from explicitly setting
+the 'user.viritofs.' path directly.
+Finally, the fourth rule lets all remaining attributes
+through.

So If I don't specify third rule, and client does setxattr(user.virtiofs.*), it will simply be a passthrough?

Quoted 4 lines

Right; and that's dangerous, because a non-privileged guest process can set a user. xattr; so a non-priv guest process could set user.virtiofs.trusted.foo and then it would get read back and be used as trusted.foo that has an impact on priviliged processes.

Quoted 7 lines

Right. We don't want unpriviliged process to be able to setup user.virtiofs.trusted.*. But that's what precisely happen in a nested configuration.

In above example, L2 will set trusted.foo, virtiofsd1 wil convert it to user.virtiofs.trusted.foo and virtiofsd0 will reject it, breaking the nested virtiofs.

So to allow nesting you need to nest the user.virtiofs. as well, not just the trusted. So either you do an all, or if you want to be more selective then I think the following would work:

Quoted 8 lines
Quoted 2 lines

1 /prefix/client/trusted./user.virtiofs./
2 /prefix/client/user.virtiofs./user.virtiofs./

Ok, so basically instead of blocking user.virtiofs.trusted. from client, prefix it with "user.virtiofs." one more time. IOW, allow client to set user.virtiofs.trusted. because it will get back user.virtiofs.trusted. and not "trusted." which is ok. Now client user space can't fool client kernel with setting arbitrary user.virtiofs.trusted xattrs.

Right.

Quoted 1 line

And if client kernel sends, trusted., it will get back trusted.

Right.

Quoted 6 lines

Only thing which can happen is that client1 sets user.virtiofs.trusted. and nested client2 will get it as trusted. So client1 user space can fool nested client's kernel. But given client1 has launched nested client2, we should be able to trust some user on client1 and make sure other users can't see this shared dir and this probably is not an issue.

Yes, that does depend a bit on how you're intending to share your filesystems etc

Quoted 12 lines
Quoted 12 lines
 3  /prefix/server//user.virtiofs./
 4  /bad/server//trusted./
 5  /ok/all///

1 causes any getattr/setattr to convert 'trusted.'
to 'user.virtiofs.trusted.'
2 causes any getattr/setattr to convert 'user.virtiofs.'
to 'user.virtiofs.user.virtiofs.'
3 causes any listattr to lose a layer of user.virtiofs. 4 blocks any trusted. from the layer beneath
5 lets anything else through

Quoted 6 lines
Quoted 3 lines

(I'm trying to convince myself if we need a /bad/server//user.virtiofs.trusted. to stop the previous level being visible).

user.virtiofs.trusted on server will be converted to trusted., right? Can't block it otherwise L1 client breaks, isn't it?

True.

Dave

Quoted 1 line

Vivek

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#27 lore raw

PATCH v3 5/5 tools/virtiofsd: xattr name mappings: Simple 'map'

In reply to Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

The mapping rule system implemented in the last few patches is extremely flexible, but not easy to use. Add a simple 'map' type as a sprinkling of sugar to make it easy.

e.g.

-o xattrmap=":map::user.virtiofs.:"

would be sufficient to prefix all xattr's
or

-o xattrmap=":map:trusted.:user.virtiofs.:"

would just prefix 'trusted.' xattr's and leave
everything else alone.

---
 docs/tools/virtiofsd.rst         |  18 ++++++
 tools/virtiofsd/passthrough_ll.c | 108 ++++++++++++++++++++++++++++++-
 2 files changed, 125 insertions(+), 1 deletion(-)
docs/tools/virtiofsd.rst +18 -0
index 5cb64612ed..e388ef253e 100644
@@ -127,6 +127,7 @@ Each rule consists of a number of fields separated with a separator that is the
 first non-white space character in the rule.  This separator must then be used
 for the whole rule.
 White space may be added before and after each rule.
+
 Using ':' as the separator a rule is of the form:
 
 ``:type:scope:key:prepend:``
@@ -162,6 +163,13 @@ on the server, and used as a new prefix.  It may be empty
 in which case a 'server' rule will always match on all names from
 the server.
 
+A simpler 'map' type provides a shorter syntax for the common case:
+
+``:map:key:prepend:``
+
+The 'map' type adds a number of separate rules to add **prepend** as a prefix
+to the matched **key** (or all attributes if **key** is empty).
+There may be at most one 'map' rule and it must be the last rule in the set.
 
 xattr-mapping Examples
 ----------------------
@@ -178,6 +186,11 @@ the first rule prefixes and strips 'user.virtiofs.',
 the second rule hides any non-prefixed attributes that
 the host set.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap=":map::user.virtiofs.:"
+
 2) Prefix 'trusted.' attributes, allow others through
 
 ::
@@ -200,6 +213,11 @@ the 'user.viritofs.' path directly.
 Finally, the fourth rule lets all remaining attributes
 through.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap="/map/trusted./user.virtiofs./"
+
 3) Hide 'security.' attributes, and allow everything else
 
 ::
tools/virtiofsd/passthrough_ll.c +107 -1

123 lines not on this page · raw message

Signature · 1 line
2.28.0
#28 lore raw

Re: [PATCH v3 5/5] tools/virtiofsd: xattr name mappings: Simple 'map'

tools/virtiofsd/passthrough_ll.c static void free_xattrmap(XattrMapEntry *map) · line 2099

Quoted 12 lines · Dr. David Alan Gilbert (git)
tools/virtiofsd/passthrough_ll.c +12 -0
+static XattrMapEntry *parse_xattrmap_map(const char *rule,
+                                         XattrMapEntry *map,
+                                         size_t *nentries)
+{
+    char sep = *rule++;
+    const char *tmp;
+    char *key;
+    char *prefix;
+    XattrMapEntry tmp_entry;
+
+    /* At start of 'key' field */
+    tmp = strchr(rule, sep);

Missing sep == '\0' check. The strchr(3) man page says:

The terminating null byte is considered part of the string, so that if c is specified as '\0', these functions return a pointer to the terminator.

So the code in this patch will eventually access beyond the end of the string:

rule = tmp + 1; <-- tmp is already at the NUL terminator

#29 lore raw

Re: [PATCH v3 5/5] tools/virtiofsd: xattr name mappings: Simple 'map'

Quoted 16 lines · * Stefan Hajnoczi ([email protected]) wrote:

tools/virtiofsd/passthrough_ll.c static void free_xattrmap(XattrMapEntry *map) · line 2099

Quoted 12 lines · Dr. David Alan Gilbert (git)
tools/virtiofsd/passthrough_ll.c +12 -0
+static XattrMapEntry *parse_xattrmap_map(const char *rule,
+                                         XattrMapEntry *map,
+                                         size_t *nentries)
+{
+    char sep = *rule++;
+    const char *tmp;
+    char *key;
+    char *prefix;
+    XattrMapEntry tmp_entry;
+
+    /* At start of 'key' field */
+    tmp = strchr(rule, sep);

Missing sep == '\0' check. The strchr(3) man page says:

Quoted 8 lines

The terminating null byte is considered part of the string, so that if c is specified as '\0', these functions return a pointer to the terminator.

So the code in this patch will eventually access beyond the end of the string:

rule = tmp + 1; <-- tmp is already at the NUL terminator

Ah yes, I've got the check in the main routine but not in the _map case; OK I'll fix that up.

Dave

Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#30 lore raw

Re: [PATCH v3 5/5] tools/virtiofsd: xattr name mappings: Simple 'map'

In reply to Dr. David Alan Gilbert (git)
Quoted 10 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

The mapping rule system implemented in the last few patches is extremely flexible, but not easy to use. Add a simple 'map' type as a sprinkling of sugar to make it easy.

e.g.

-o xattrmap=":map::user.virtiofs.:"

Quoted 7 lines

would be sufficient to prefix all xattr's
or

-o xattrmap=":map:trusted.:user.virtiofs.:"

would just prefix 'trusted.' xattr's and leave
everything else alone.

Will it block "user.virtiofs.trusted." from client? As we discussed that either we need to block it or we need to prefix it with another user.virtiofs. I mean this rule alone is problematic and needs to be coupled with more rules.

I am assuming one can specify multiple xattrmap on single line. So one can also say.

-o xattrmap=":map:trusted.:user.virtiofs.:" -o xattrmap=":map:user.virtiofs.:user.virtiofs."

Thanks
Vivek

Quoted 189 lines
---
 docs/tools/virtiofsd.rst         |  18 ++++++
 tools/virtiofsd/passthrough_ll.c | 108 ++++++++++++++++++++++++++++++-
 2 files changed, 125 insertions(+), 1 deletion(-)
docs/tools/virtiofsd.rst +18 -0
index 5cb64612ed..e388ef253e 100644
@@ -127,6 +127,7 @@ Each rule consists of a number of fields separated with a separator that is the
 first non-white space character in the rule.  This separator must then be used
 for the whole rule.
 White space may be added before and after each rule.
+
 Using ':' as the separator a rule is of the form:
 
 ``:type:scope:key:prepend:``
@@ -162,6 +163,13 @@ on the server, and used as a new prefix.  It may be empty
 in which case a 'server' rule will always match on all names from
 the server.
 
+A simpler 'map' type provides a shorter syntax for the common case:
+
+``:map:key:prepend:``
+
+The 'map' type adds a number of separate rules to add **prepend** as a prefix
+to the matched **key** (or all attributes if **key** is empty).
+There may be at most one 'map' rule and it must be the last rule in the set.
 
 xattr-mapping Examples
 ----------------------
@@ -178,6 +186,11 @@ the first rule prefixes and strips 'user.virtiofs.',
 the second rule hides any non-prefixed attributes that
 the host set.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap=":map::user.virtiofs.:"
+
 2) Prefix 'trusted.' attributes, allow others through
 
 ::
@@ -200,6 +213,11 @@ the 'user.viritofs.' path directly.
 Finally, the fourth rule lets all remaining attributes
 through.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap="/map/trusted./user.virtiofs./"
+
 3) Hide 'security.' attributes, and allow everything else
 
 ::
tools/virtiofsd/passthrough_ll.c +107 -1
index 8406a2ae86..a1b3364ba3 100644
@@ -2074,6 +2074,106 @@ static void free_xattrmap(XattrMapEntry *map)
     g_free(map);
 }
 
+/*
+ * Handle the 'map' type, which is sugar for a set of commands
+ * for the common case of prefixing a subset or everything,
+ * and allowing anything not prefixed through.
+ * It must be the last entry in the stream, although there
+ * can be other entries before it.
+ * The form is:
+ *    :map:key:prefix:
+ *
+ * key maybe empty in which case all entries are prefixed.
+ */
+static XattrMapEntry *parse_xattrmap_map(const char *rule,
+                                         XattrMapEntry *map,
+                                         size_t *nentries)
+{
+    char sep = *rule++;
+    const char *tmp;
+    char *key;
+    char *prefix;
+    XattrMapEntry tmp_entry;
+
+    /* At start of 'key' field */
+    tmp = strchr(rule, sep);
+    if (!tmp) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Missing '%c' at end of key field in map rule\n",
+                 __func__, sep);
+        exit(1);
+    }
+
+    key = g_strndup(rule, tmp - rule);
+    rule = tmp + 1;
+
+    /* At start of prefix field */
+    tmp = strchr(rule, sep);
+    if (!tmp) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Missing '%c' at end of prefix field in map rule\n",
+                 __func__, sep);
+        exit(1);
+    }
+
+    prefix = g_strndup(rule, tmp - rule);
+    rule = tmp + 1;
+
+    /*
+     * This should be the end of the string, we don't allow
+     * any more commands after 'map'.
+     */
+    if (*rule) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Expecting end of command after map, found '%c'\n",
+                 __func__, *rule);
+        exit(1);
+    }
+
+    /* 1st: Prefix matches/everything */
+    tmp_entry.flags = XATTR_MAP_FLAG_PREFIX | XATTR_MAP_FLAG_ALL;
+    tmp_entry.key = g_strdup(key);
+    tmp_entry.prepend = g_strdup(prefix);
+    map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+    if (!*key) {
+        /* Prefix all case */
+
+        /* 2nd: Hide any non-prefixed entries on the host */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_ALL |
+                          XATTR_MAP_FLAG_LAST;
+        tmp_entry.key = g_strdup("");
+        tmp_entry.prepend = g_strdup("");
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+    } else {
+        /* Prefix matching case */
+
+        /* 2nd: Hide non-prefixed but matching entries on the host */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_SERVER;
+        tmp_entry.key = g_strdup(""); /* Not used */
+        tmp_entry.prepend = g_strdup(key);
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+        /* 3rd: Stop the client accessing prefixed attributes directly */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_CLIENT;
+        tmp_entry.key = g_strdup(prefix);
+        tmp_entry.prepend = g_strdup(""); /* Not used */
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+        /* 4th: Everything else is OK */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_OK | XATTR_MAP_FLAG_ALL |
+                          XATTR_MAP_FLAG_LAST;
+        tmp_entry.key = g_strdup("");
+        tmp_entry.prepend = g_strdup("");
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+    }
+
+    g_free(key);
+    g_free(prefix);
+
+    return map;
+}
+
 static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
 {
     XattrMapEntry *res = NULL;
@@ -2102,10 +2202,16 @@ static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
             tmp_entry.flags |= XATTR_MAP_FLAG_END_OK;
         } else if (strstart(map, "bad", &map)) {
             tmp_entry.flags |= XATTR_MAP_FLAG_END_BAD;
+        } else if (strstart(map, "map", &map)) {
+            /*
+             * map is sugar that adds a number of rules, and must be
+             * the last entry.
+             */
+            return parse_xattrmap_map(map, res, &nentries);
         } else {
             fuse_log(FUSE_LOG_ERR,
                      "%s: Unexpected type;"
-                     "Expecting 'prefix', 'ok', or 'bad' in rule %zu\n",
+                     "Expecting 'prefix', 'ok', 'bad' or 'map' in rule %zu\n",
                      __func__, nentries);
             exit(1);
         }
Signature · 1 line
2.28.0
#31 lore raw

Re: [PATCH v3 5/5] tools/virtiofsd: xattr name mappings: Simple 'map'

Quoted 16 lines · * Vivek Goyal ([email protected]) wrote:
Quoted 15 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

The mapping rule system implemented in the last few patches is extremely flexible, but not easy to use. Add a simple 'map' type as a sprinkling of sugar to make it easy.

e.g.

-o xattrmap=":map::user.virtiofs.:"

would be sufficient to prefix all xattr's
or

-o xattrmap=":map:trusted.:user.virtiofs.:"

Quoted 7 lines
Quoted 2 lines

would just prefix 'trusted.' xattr's and leave
everything else alone.

Will it block "user.virtiofs.trusted." from client? As we discussed that either we need to block it or we need to prefix it with another user.virtiofs. I mean this rule alone is problematic and needs to be coupled with more rules.

It blocked user.virtiofs. more generically:

+        /* 2nd: Hide non-prefixed but matching entries on the host */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_SERVER;
+        tmp_entry.key = g_strdup(""); /* Not used */
+        tmp_entry.prepend = g_strdup(key);
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);

i.e. hides 'trusted.'

+        /* 3rd: Stop the client accessing prefixed attributes directly */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_CLIENT;
+        tmp_entry.key = g_strdup(prefix);
+        tmp_entry.prepend = g_strdup(""); /* Not used */
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);

i.e. blocks client from 'user.virtiofs.'

Note, that is a problem for nesting, since having that at L1 means it stops L2 from settings user.virtiofs.trusted.something (which it wants to become user.virtiofs.user.virtiofs.trusted.something).

Using :map::user.virtiofs.: works nicely for nesting (I've just tried it), since that nests everything.

Quoted 4 lines

I am assuming one can specify multiple xattrmap on single line. So one can also say.

-o xattrmap=":map:trusted.:user.virtiofs.:" -o xattrmap=":map:user.virtiofs.:user.virtiofs."

No, the existing core option parser we're using overrides on subsequent entries; so that would ignore the first -o xattrmap=

Dave

Quoted 193 lines

Thanks
Vivek

Quoted 189 lines
---
 docs/tools/virtiofsd.rst         |  18 ++++++
 tools/virtiofsd/passthrough_ll.c | 108 ++++++++++++++++++++++++++++++-
 2 files changed, 125 insertions(+), 1 deletion(-)
docs/tools/virtiofsd.rst +18 -0
index 5cb64612ed..e388ef253e 100644
@@ -127,6 +127,7 @@ Each rule consists of a number of fields separated with a separator that is the
 first non-white space character in the rule.  This separator must then be used
 for the whole rule.
 White space may be added before and after each rule.
+
 Using ':' as the separator a rule is of the form:
 
 ``:type:scope:key:prepend:``
@@ -162,6 +163,13 @@ on the server, and used as a new prefix.  It may be empty
 in which case a 'server' rule will always match on all names from
 the server.
 
+A simpler 'map' type provides a shorter syntax for the common case:
+
+``:map:key:prepend:``
+
+The 'map' type adds a number of separate rules to add **prepend** as a prefix
+to the matched **key** (or all attributes if **key** is empty).
+There may be at most one 'map' rule and it must be the last rule in the set.
 
 xattr-mapping Examples
 ----------------------
@@ -178,6 +186,11 @@ the first rule prefixes and strips 'user.virtiofs.',
 the second rule hides any non-prefixed attributes that
 the host set.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap=":map::user.virtiofs.:"
+
 2) Prefix 'trusted.' attributes, allow others through
 
 ::
@@ -200,6 +213,11 @@ the 'user.viritofs.' path directly.
 Finally, the fourth rule lets all remaining attributes
 through.
 
+This is equivalent to the 'map' rule:
+
+::
+-o xattrmap="/map/trusted./user.virtiofs./"
+
 3) Hide 'security.' attributes, and allow everything else
 
 ::
tools/virtiofsd/passthrough_ll.c +107 -1
index 8406a2ae86..a1b3364ba3 100644
@@ -2074,6 +2074,106 @@ static void free_xattrmap(XattrMapEntry *map)
     g_free(map);
 }
 
+/*
+ * Handle the 'map' type, which is sugar for a set of commands
+ * for the common case of prefixing a subset or everything,
+ * and allowing anything not prefixed through.
+ * It must be the last entry in the stream, although there
+ * can be other entries before it.
+ * The form is:
+ *    :map:key:prefix:
+ *
+ * key maybe empty in which case all entries are prefixed.
+ */
+static XattrMapEntry *parse_xattrmap_map(const char *rule,
+                                         XattrMapEntry *map,
+                                         size_t *nentries)
+{
+    char sep = *rule++;
+    const char *tmp;
+    char *key;
+    char *prefix;
+    XattrMapEntry tmp_entry;
+
+    /* At start of 'key' field */
+    tmp = strchr(rule, sep);
+    if (!tmp) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Missing '%c' at end of key field in map rule\n",
+                 __func__, sep);
+        exit(1);
+    }
+
+    key = g_strndup(rule, tmp - rule);
+    rule = tmp + 1;
+
+    /* At start of prefix field */
+    tmp = strchr(rule, sep);
+    if (!tmp) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Missing '%c' at end of prefix field in map rule\n",
+                 __func__, sep);
+        exit(1);
+    }
+
+    prefix = g_strndup(rule, tmp - rule);
+    rule = tmp + 1;
+
+    /*
+     * This should be the end of the string, we don't allow
+     * any more commands after 'map'.
+     */
+    if (*rule) {
+        fuse_log(FUSE_LOG_ERR,
+                 "%s: Expecting end of command after map, found '%c'\n",
+                 __func__, *rule);
+        exit(1);
+    }
+
+    /* 1st: Prefix matches/everything */
+    tmp_entry.flags = XATTR_MAP_FLAG_PREFIX | XATTR_MAP_FLAG_ALL;
+    tmp_entry.key = g_strdup(key);
+    tmp_entry.prepend = g_strdup(prefix);
+    map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+    if (!*key) {
+        /* Prefix all case */
+
+        /* 2nd: Hide any non-prefixed entries on the host */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_ALL |
+                          XATTR_MAP_FLAG_LAST;
+        tmp_entry.key = g_strdup("");
+        tmp_entry.prepend = g_strdup("");
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+    } else {
+        /* Prefix matching case */
+
+        /* 2nd: Hide non-prefixed but matching entries on the host */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_SERVER;
+        tmp_entry.key = g_strdup(""); /* Not used */
+        tmp_entry.prepend = g_strdup(key);
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+        /* 3rd: Stop the client accessing prefixed attributes directly */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_BAD | XATTR_MAP_FLAG_CLIENT;
+        tmp_entry.key = g_strdup(prefix);
+        tmp_entry.prepend = g_strdup(""); /* Not used */
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+
+        /* 4th: Everything else is OK */
+        tmp_entry.flags = XATTR_MAP_FLAG_END_OK | XATTR_MAP_FLAG_ALL |
+                          XATTR_MAP_FLAG_LAST;
+        tmp_entry.key = g_strdup("");
+        tmp_entry.prepend = g_strdup("");
+        map = add_xattrmap_entry(map, nentries, &tmp_entry);
+    }
+
+    g_free(key);
+    g_free(prefix);
+
+    return map;
+}
+
 static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
 {
     XattrMapEntry *res = NULL;
@@ -2102,10 +2202,16 @@ static XattrMapEntry *parse_xattrmap(struct lo_data *lo)
             tmp_entry.flags |= XATTR_MAP_FLAG_END_OK;
         } else if (strstart(map, "bad", &map)) {
             tmp_entry.flags |= XATTR_MAP_FLAG_END_BAD;
+        } else if (strstart(map, "map", &map)) {
+            /*
+             * map is sugar that adds a number of rules, and must be
+             * the last entry.
+             */
+            return parse_xattrmap_map(map, res, &nentries);
         } else {
             fuse_log(FUSE_LOG_ERR,
                      "%s: Unexpected type;"
-                     "Expecting 'prefix', 'ok', or 'bad' in rule %zu\n",
+                     "Expecting 'prefix', 'ok', 'bad' or 'map' in rule %zu\n",
                      __func__, nentries);
             exit(1);
         }
Signature · 1 line
2.28.0
Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK