git:master: modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned
Module: sip-router Branch: master Commit: f15de29015546d173a1d3135f90653e05d85171f URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=f15de290...
Author: Jason Penton jason.penton@gmail.com Committer: Jason Penton jason.penton@gmail.com Date: Mon Mar 10 14:38:51 2014 +0200
modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned - related to earlier commit 4ab0f53ff247f411dde7b88d5b7d82fc7e17baa9
---
modules/corex/corex_lib.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/modules/corex/corex_lib.c b/modules/corex/corex_lib.c index 9d926c2..f83352b 100644 --- a/modules/corex/corex_lib.c +++ b/modules/corex/corex_lib.c @@ -84,7 +84,8 @@ int corex_append_branch(sip_msg_t *msg, gparam_t *pu, gparam_t *pq) msg->dst_uri.s = 0; msg->dst_uri.len = 0; if(msg->path_vec.s!=0) - pkg_free(msg->path_vec.s); + if (likely(msg->path_vec.s >= msg->buf && (msg->path_vec.s < (msg->buf + msg->len)))) + pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0; }
Hello,
the condition to do pkg_free() or not for msg->path_vec.s doesn't seem correct. You check if path_vec.s points inside SIP message buffer, but that doesn't happen at all. Path vector is taken from usrloc record in cloned in pkg for sip_msg structure.
Did I misunderstand something?
Cheers, Daniel
On 10/03/14 13:39, Jason Penton wrote:
Module: sip-router Branch: master Commit: f15de29015546d173a1d3135f90653e05d85171f URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=f15de290...
Author: Jason Penton jason.penton@gmail.com Committer: Jason Penton jason.penton@gmail.com Date: Mon Mar 10 14:38:51 2014 +0200
modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned
- related to earlier commit 4ab0f53ff247f411dde7b88d5b7d82fc7e17baa9
modules/corex/corex_lib.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/modules/corex/corex_lib.c b/modules/corex/corex_lib.c index 9d926c2..f83352b 100644 --- a/modules/corex/corex_lib.c +++ b/modules/corex/corex_lib.c @@ -84,7 +84,8 @@ int corex_append_branch(sip_msg_t *msg, gparam_t *pu, gparam_t *pq) msg->dst_uri.s = 0; msg->dst_uri.len = 0; if(msg->path_vec.s!=0)
pkg_free(msg->path_vec.s);
if (likely(msg->path_vec.s >= msg->buf && (msg->path_vec.s < (msg->buf + msg->len))))pkg_free(msg->path_vec.s);
sr-dev mailing list sr-dev@lists.sip-router.org http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev
Hey Daniel,
You are correct. Apologies, can you revert that commit and I will look into it. While this works for not freeing pkg mem on a shm clone - I see it will break in the other case. I suspect the only way to do it now is to use msg flags?
Cheers Jason
On Mon, Mar 10, 2014 at 3:31 PM, Daniel-Constantin Mierla <miconda@gmail.com
wrote:
Hello,
the condition to do pkg_free() or not for msg->path_vec.s doesn't seem correct. You check if path_vec.s points inside SIP message buffer, but that doesn't happen at all. Path vector is taken from usrloc record in cloned in pkg for sip_msg structure.
Did I misunderstand something?
Cheers, Daniel
On 10/03/14 13:39, Jason Penton wrote:
Module: sip-router Branch: master Commit: f15de29015546d173a1d3135f90653e05d85171f URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a= commit;h=f15de29015546d173a1d3135f90653e05d85171f
Author: Jason Penton jason.penton@gmail.com Committer: Jason Penton jason.penton@gmail.com Date: Mon Mar 10 14:38:51 2014 +0200
modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned - related to earlier commit 4ab0f53ff247f411dde7b88d5b7d82 fc7e17baa9
modules/corex/corex_lib.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/modules/corex/corex_lib.c b/modules/corex/corex_lib.c index 9d926c2..f83352b 100644 --- a/modules/corex/corex_lib.c +++ b/modules/corex/corex_lib.c @@ -84,7 +84,8 @@ int corex_append_branch(sip_msg_t *msg, gparam_t *pu, gparam_t *pq) msg->dst_uri.s = 0; msg->dst_uri.len = 0; if(msg->path_vec.s!=0)
pkg_free(msg->path_vec.s);
if (likely(msg->path_vec.s >= msg->buf &&(msg->path_vec.s < (msg->buf + msg->len))))
pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0; }
sr-dev mailing list sr-dev@lists.sip-router.org http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev
-- Daniel-Constantin Mierla - http://www.asipto.com http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda Kamailio World Conference - April 2-4, 2014, Berlin, Germany http://www.kamailioworld.com
Hello,
using internal msg flags is one solution for now. Probably you can push quickly a patch starting from current code, instead of reverting -- there were few related/similar commits, as I could spot.
When I get a chance I will push a function that can tell if a pointer is in shared memory. Still the flag could be safer to know if the value has to be freed (pkg/shm) or not at all.
Cheers, Daniel
On 10/03/14 14:34, Jason Penton wrote:
Hey Daniel,
You are correct. Apologies, can you revert that commit and I will look into it. While this works for not freeing pkg mem on a shm clone - I see it will break in the other case. I suspect the only way to do it now is to use msg flags?
Cheers Jason
On Mon, Mar 10, 2014 at 3:31 PM, Daniel-Constantin Mierla <miconda@gmail.com mailto:miconda@gmail.com> wrote:
Hello, the condition to do pkg_free() or not for msg->path_vec.s doesn't seem correct. You check if path_vec.s points inside SIP message buffer, but that doesn't happen at all. Path vector is taken from usrloc record in cloned in pkg for sip_msg structure. Did I misunderstand something? Cheers, Daniel On 10/03/14 13:39, Jason Penton wrote: Module: sip-router Branch: master Commit: f15de29015546d173a1d3135f90653e05d85171f URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=f15de29015546d173a1d3135f90653e05d85171f Author: Jason Penton <jason.penton@gmail.com <mailto:jason.penton@gmail.com>> Committer: Jason Penton <jason.penton@gmail.com <mailto:jason.penton@gmail.com>> Date: Mon Mar 10 14:38:51 2014 +0200 modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned - related to earlier commit 4ab0f53ff247f411dde7b88d5b7d82fc7e17baa9 --- modules/corex/corex_lib.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/modules/corex/corex_lib.c b/modules/corex/corex_lib.c index 9d926c2..f83352b 100644 --- a/modules/corex/corex_lib.c +++ b/modules/corex/corex_lib.c @@ -84,7 +84,8 @@ int corex_append_branch(sip_msg_t *msg, gparam_t *pu, gparam_t *pq) msg->dst_uri.s = 0; msg->dst_uri.len = 0; if(msg->path_vec.s!=0) - pkg_free(msg->path_vec.s); + if (likely(msg->path_vec.s >= msg->buf && (msg->path_vec.s < (msg->buf + msg->len)))) + pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0; } _______________________________________________ sr-dev mailing list sr-dev@lists.sip-router.org <mailto:sr-dev@lists.sip-router.org> http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev -- Daniel-Constantin Mierla - http://www.asipto.com http://twitter.com/#!/miconda <http://twitter.com/#%21/miconda> - http://www.linkedin.com/in/miconda Kamailio World Conference - April 2-4, 2014, Berlin, Germany http://www.kamailioworld.com
Sure Daniel, I'm driving now but will patch as soon as I get home
-- Sent using my phone and may be brief, to the point or contain typos -- On Mar 10, 2014 3:43 PM, "Daniel-Constantin Mierla" miconda@gmail.com wrote:
Hello,
using internal msg flags is one solution for now. Probably you can push quickly a patch starting from current code, instead of reverting -- there were few related/similar commits, as I could spot.
When I get a chance I will push a function that can tell if a pointer is in shared memory. Still the flag could be safer to know if the value has to be freed (pkg/shm) or not at all.
Cheers, Daniel
On 10/03/14 14:34, Jason Penton wrote:
Hey Daniel,
You are correct. Apologies, can you revert that commit and I will look into it. While this works for not freeing pkg mem on a shm clone - I see it will break in the other case. I suspect the only way to do it now is to use msg flags?
Cheers Jason
On Mon, Mar 10, 2014 at 3:31 PM, Daniel-Constantin Mierla < miconda@gmail.com> wrote:
Hello,
the condition to do pkg_free() or not for msg->path_vec.s doesn't seem correct. You check if path_vec.s points inside SIP message buffer, but that doesn't happen at all. Path vector is taken from usrloc record in cloned in pkg for sip_msg structure.
Did I misunderstand something?
Cheers, Daniel
On 10/03/14 13:39, Jason Penton wrote:
Module: sip-router Branch: master Commit: f15de29015546d173a1d3135f90653e05d85171f URL: http://git.sip-router.org/cgi-bin/gitweb.cgi/sip-router/?a=commit;h=f15de290...
Author: Jason Penton jason.penton@gmail.com Committer: Jason Penton jason.penton@gmail.com Date: Mon Mar 10 14:38:51 2014 +0200
modules/corex: only free path_vector from pkg if it is still in pkg... not shm cloned - related to earlier commit 4ab0f53ff247f411dde7b88d5b7d82fc7e17baa9
modules/corex/corex_lib.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/modules/corex/corex_lib.c b/modules/corex/corex_lib.c index 9d926c2..f83352b 100644 --- a/modules/corex/corex_lib.c +++ b/modules/corex/corex_lib.c @@ -84,7 +84,8 @@ int corex_append_branch(sip_msg_t *msg, gparam_t *pu, gparam_t *pq) msg->dst_uri.s = 0; msg->dst_uri.len = 0; if(msg->path_vec.s!=0)
pkg_free(msg->path_vec.s);
if (likely(msg->path_vec.s >= msg->buf &&(msg->path_vec.s < (msg->buf + msg->len))))
pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0; }
sr-dev mailing list sr-dev@lists.sip-router.org http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev
-- Daniel-Constantin Mierla - http://www.asipto.com http://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda Kamailio World Conference - April 2-4, 2014, Berlin, Germany http://www.kamailioworld.com
-- Daniel-Constantin Mierla - http://www.asipto.comhttp://twitter.com/#!/miconda - http://www.linkedin.com/in/miconda Kamailio World Conference - April 2-4, 2014, Berlin, Germanyhttp://www.kamailioworld.com
-
Daniel-Constantin Mierla -
Jason Penton