[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <4B0A5FC4.1010902@canonical.com>
Date: Mon, 23 Nov 2009 02:11:16 -0800
From: John Johansen <john.johansen@...onical.com>
To: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
CC: linux-kernel@...r.kernel.org, linux-security-module@...r.kernel.org
Subject: Re: [AppArmor #3 0/12] AppArmor security module
Tetsuo Handa wrote:
> Regarding file.c ipc.c lib.c lsm.c
>
>
>
> You can use container_of() inside callback functions to avoid "void *".
>
yeah that is cleaner, will do
>> int aa_audit(int type, struct aa_profile *profile, struct aa_audit *sa,
>> void (*cb) (struct audit_buffer *, void *))
>
> int aa_audit(int type, struct aa_profile *profile, struct aa_audit *sa,
> void (*cb) (struct audit_buffer *, struct aa_audit *))
>
>> static int aa_audit_base(int type, struct aa_profile *profile,
>> struct aa_audit *sa, struct audit_context *audit_cxt,
>> void (*cb) (struct audit_buffer *, void *))
>
> static int aa_audit_base(int type, struct aa_profile *profile,
> struct aa_audit *sa, struct audit_context *audit_cxt,
> void (*cb) (struct audit_buffer *, struct aa_audit *))
>
>> void file_audit_cb(struct audit_buffer *ab, void *va)
>> {
>> struct aa_audit_file *sa = va;
>
> void file_audit_cb(struct audit_buffer *ab, struct aa_audit *va)
> {
> struct aa_audit_file *sa = container_of(va, struct aa_audit_file, base);
>
>> int aa_audit_file(struct aa_profile *profile, struct aa_audit_file *sa)
>> (...snipped...)
>> return aa_audit(type, profile, (struct aa_audit *)sa, file_audit_cb);
>
> return aa_audit(type, profile, &sa->base, file_audit_cb);
>
>> }
>
>> static void audit_cb(struct audit_buffer *ab, void *va)
>> {
>> struct aa_audit_ptrace *sa = va;
>
> static void audit_cb(struct audit_buffer *ab, struct aa_audit *va)
> {
> struct aa_audit_ptrace *sa = container_of(va, struct aa_audit_ptrace,
> base);
>
>> static int aa_audit_ptrace(struct aa_profile *profile,
>> struct aa_audit_ptrace *sa)
>> {
>> return aa_audit(AUDIT_APPARMOR_AUTO, profile, (struct aa_audit *)sa,
>> audit_cb);
>
> return aa_audit(AUDIT_APPARMOR_AUTO, profile, &sa->base, audit_cb);
>
>
>
>> int aa_path_link(struct aa_profile *profile, struct dentry *old_dentry,
>> struct path *new_dir, struct dentry *new_dentry)
> (.,..snipped...)
>> unsigned int state;
> (.,..snipped...)
>> sa.perms = aa_str_perms(profile->file.dfa, DFA_START, sa.name, &cond,
>> &state);
>
> "state" remains uninitialized if profile->file.dfa == NULL.
> Are you sure profile->file.dfa != NULL ?
>
No this needs to be fixed.
>
>
>> char *aa_strchrnul(const char *s, int c)
>> {
>> for (; *s != (char)c && *s != '\0'; ++s)
>> ;
>> return (char *)s;
>> }
>
> Only fqname_subname() calls aa_strchrnul() and
> fqname_subname() returns NULL if aa_strchrnul() returns '\0'.
> You can use strchr() instead.
>
hrmm right thanks
>> static const char *fqname_subname(const char *name)
>> {
>> char *split;
>> /* check for namespace which begins with a : and ends with : or \0 */
>> name = strstrip((char *)name);
>> if (*name == ':') {
>> split = aa_strchrnul(name + 1, ':');
>> if (*split == '\0')
>> return NULL;
>
> split = strchr(name + 1, ':');
> if (!split)
> return NULL;
>
yep
>> name = strstrip(split + 1);
>> }
>> for (split = strstr(name, "//"); split; split = strstr(name, "//"))
>> name = split + 2;
>>
>> return name;
>> }
>
>
>
>> char *aa_split_name_from_ns(char *args, char **ns_name)
>> {
>> char *name = strstrip(args);
>>
>> *ns_name = NULL;
>> if (args[0] == ':') {
>> char *split = strstrip(strchr(&args[1], ':'));
>>
>> if (!split)
>> return NULL;
>
>
> strchr() returns NULL if not found, and strstrip(NULL) will do strlen(NULL).
> strstrip() never returns NULL. Did you mean
>
> char *split = strchr(&args[1], ':');
>
> if (!split)
> return NULL;
> split = strstrip(split);
>
> ?
yes, thanks
>
>> *split = 0;
>> *ns_name = &args[1];
>> name = strstrip(split + 1);
>> }
>> if (*name == 0)
>> name = NULL;
>>
>> return name;
>> }
>
>
>
>> static int apparmor_sysctl(struct ctl_table *table, int op)
> This hook will be removed.
>
>> char *sysctl_pathname(struct ctl_table *table, char *buffer, int buflen)
> This function will no longer be needed.
>
>> int aa_pathstr_perm(struct aa_profile *profile, const char *op,
>> const char *name, u16 request, struct path_cond *cond)
> This function will no longer be needed.
>
yep
>
>
>> static int apparmor_file_permission(struct file *file, int mask)
>> {
>> /*
>> * TODO: cache profiles that have revalidated?
>> */
>> struct aa_file_cxt *fcxt = file->f_security;
>> struct aa_profile *profile, *fprofile = fcxt->profile;
>> int error = 0;
>>
>> if (!fprofile || !file->f_path.mnt ||
>> !mediated_filesystem(file->f_path.dentry->d_inode))
>> return 0;
>>
>> profile = aa_current_profile();
>>
>> #ifdef CONFIG_SECURITY_APPARMOR_COMPAT_24
>> /*
>> * AppArmor <= 2.4 revalidates files at access time instead
>> * of at exec.
>> */
>> if (profile && ((fprofile != profile) || (mask & ~fcxt->allowed)))
>> error = aa_file_perm(profile, "file_perm", file, mask);
>> #endif
>>
>> return error;
>> }
>
> Why need to call this function if CONFIG_SECURITY_APPARMOR_COMPAT_24=n ?
> I think we can do
>
>> static struct security_operations apparmor_ops = {
> (...snipped...)
>
> #ifdef CONFIG_SECURITY_APPARMOR_COMPAT_24
>
>> .file_permission = apparmor_file_permission,
>
> #endif
>
yes we can currently. Though this will change in the future, but for now
we should got with the cleaner switch.
> (...snipped...)
>> }
>
>
>
>> int aa_alloc_default_namespace(void)
>
> This function could be declared with __init attribute.
>
yep, thanks
>
>
>> static int __init apparmor_init(void)
> (...snipped...)
>> error = set_init_cxt();
>> if (error) {
>> AA_ERROR("Failed to set context on init task\n");
>> goto alloc_out;
>
> This should be
>
> goto register_security_out;
>
> in order to call aa_free_default_namespace().
>
>> }
indeed
thanks again for taking the time to review
john
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists