Message ID | 20180516171124.24962-3-vivek@collabora.com |
---|---|
State | New |
Headers | show |
Series | Proof-of-Concept implementation of RTLD_SHARED for dlmopen | expand |
On 05/16/2018 01:11 PM, Vivek Das Mohapatra wrote: > From: Vivek Das Mohapatra <vivek@collabora.co.uk> > > We already have an l_real pointer, used for a similar purpose by > the linker for copies of ld.so in secondary namespaces. Update its > documentation and add a bitfield to indicate when link_map entry > is a clone. I would avoid calling this a "clone" because it implies a direct copy that is mutable. This is just bikeshedding but I think 'l_proxy' might be more accurate? That way we can talk about inserting a proxy object into a namespace and know that the proxy is just that, a thin proxy which points to complete link map. > --- > include/link.h | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/include/link.h b/include/link.h > index 5924594548..e051b5cd36 100644 > --- a/include/link.h > +++ b/include/link.h > @@ -104,8 +104,9 @@ struct link_map > They may change without notice. */ > > /* This is an element which is only ever different from a pointer to > - the very same copy of this type for ld.so when it is used in more > - than one namespace. */ > + the very same copy of this type when: > + - A shallow copy of ld.so is placed in namespaces other than LM_ID_BASE. > + - An object is cloned into a namespace by dlmopen with RTLD_SHARED. */ OK. > struct link_map *l_real; > > /* Number of the namespace this link map belongs to. */ > @@ -177,6 +178,7 @@ struct link_map > unsigned int l_relocated:1; /* Nonzero if object's relocations done. */ > unsigned int l_init_called:1; /* Nonzero if DT_INIT function called. */ > unsigned int l_global:1; /* Nonzero if object in _dl_global_scope. */ > + unsigned int l_clone:1; /* Nonzero if object is a clone. */ Suggest l_proxy:1; /* Nonzero if object is a proxy. */ > unsigned int l_reserved:2; /* Reserved for internal use. */ > unsigned int l_phdr_allocated:1; /* Nonzero if the data structure pointed > to by `l_phdr' is allocated. */ >
> This is just bikeshedding but I think 'l_proxy' might be more accurate? Sounds good. I initially named it clone before I'd fully dug into the implementation details. > That way we can talk about inserting a proxy object into a namespace > and know that the proxy is just that, a thin proxy which points to > complete link map. Sounds good to me.
diff --git a/include/link.h b/include/link.h index 5924594548..e051b5cd36 100644 --- a/include/link.h +++ b/include/link.h @@ -104,8 +104,9 @@ struct link_map They may change without notice. */ /* This is an element which is only ever different from a pointer to - the very same copy of this type for ld.so when it is used in more - than one namespace. */ + the very same copy of this type when: + - A shallow copy of ld.so is placed in namespaces other than LM_ID_BASE. + - An object is cloned into a namespace by dlmopen with RTLD_SHARED. */ struct link_map *l_real; /* Number of the namespace this link map belongs to. */ @@ -177,6 +178,7 @@ struct link_map unsigned int l_relocated:1; /* Nonzero if object's relocations done. */ unsigned int l_init_called:1; /* Nonzero if DT_INIT function called. */ unsigned int l_global:1; /* Nonzero if object in _dl_global_scope. */ + unsigned int l_clone:1; /* Nonzero if object is a clone. */ unsigned int l_reserved:2; /* Reserved for internal use. */ unsigned int l_phdr_allocated:1; /* Nonzero if the data structure pointed to by `l_phdr' is allocated. */
From: Vivek Das Mohapatra <vivek@collabora.co.uk> We already have an l_real pointer, used for a similar purpose by the linker for copies of ld.so in secondary namespaces. Update its documentation and add a bitfield to indicate when link_map entry is a clone. --- include/link.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)