- 19 May, 2017 12 commits
-
-
As the object referred to by sys is allocated with calloc, sys->tcp_sock will be equal to 0 if an error occurs prior to it being assigned a value in satip_open (such as when encountering a malformed URI). This will cause the error clean-up to assume that tcp->tcp_sock refers to an open file-descriptor, triggering an assert if no suck file-descriptor exists or closing a fd that access/satip definitely should not mess with. fixes: #18344 Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org> -
Jean-Baptiste Kempf authored -
p_sys->p_es and p_sys->p_pes are treated as block-chains, and not as single blocks. As such block_ChainRelease shall be used to release the full chain, as block_Release will only release the front-most block. fixes: #17997 Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
The entries within p_sys->attachments are of course allocated with vlc_input_attachment_New, meaning that they shall be released using the corresponding vlc_input_attachment_Delete. fixes: #18288 Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
François Cartegnie authored
refs #18245
-
The previous implementation would potentially try to duplicate a seekpoint that was never used due to invalid arguments, as well as leaking said duplicate. These changes make sure that we do not leak the resource, while also preventing the unnecessary duplication + returning an error if we are unable to complete the operation. CID: 1409711 Signed-off-by:Thomas Guillem <thomas@gllm.fr>
-
used in I420 to NV12 conversion Signed-off-by: -
Signed-off-by: -
Without this it's impossible to do some processing like splitting NV12 UV planes, it requires double the resolution. In that case the code asserts on assert(hstep > 0) in SSE_SplitPlanes(). Signed-off-by: -
Signed-off-by: -
Thomas Guillem authored
This fixes a regression introduced by f9f0bbc1 This memory corruption was not visible since this function is generally used to copy from GPU mapped buffers, that are often bigger than what you ask.
-
The variable for the current index is named "row", and not "i". Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org>
-
- 18 May, 2017 28 commits
-
-
If UpdateBookmarksOption is called more than once, where the generated bookmark-variable string would be different, it will result in the input_item_t having several entries for "bookmarks=". This patch makes it so that we overwrite any existing entry if such is available, instead of simply appending a new one. fixes: #18338 Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Writing to killed needs to be protected by the mutex, there is also a possibility of losing the sent signal unless we do it while the mutex is acquired (a signal is not a semaphore). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
There is absolutely no need to lock and unlock the mutex on every iteration, especially given that one cannot observe this behavior (as there is nothing at all that guarantees that something can aquire the lock inbetween those two calls). There is also no need to check if we are killed twice, once is sufficient Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Due to the changes introduced by 36e12797 , the bookmark menu would not be enabled if there were no choices currently associated with the input. As a user should be able to manage the bookmarks, even if no such is present at the current time, we forcefully add VLC_VAR_HASCHOICE to the i_type. Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
This is the same bug as was recently fixed by 1cf21dd4 , namely that the individual resources within the two resulting arrays were not released (causing a leak). Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
vlc_dictionary_all_keys may return NULL on error, rendering the previous implemention prone to a potential null-pointer dereference. Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
vlc_dictionary_all_keys returns NULL on error, meaning that the previous implementation could result in a null-pointer dereference. Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
The changes introduced in 08a2eb61 correctly adds a new block for reading from p_extra, it however left out the if-condition to protect said block. Signed-off-by:
Jean-Baptiste Kempf <jb@videolan.org> -
There is no need to use reinterpret_cast when the source or destination type is pointer-to-void, static_cast is more accurate (and safer). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
There is no need to use reinterpret_cast when the source or destination type is pointer-to-void, static_cast is more accurate (and safer). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
There is no need to use reinterpret_cast when the destination or source type is pointer-to-void, static_cast is more accurate (and safer). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
There is no need to use reinterpret_cast when the destination or source type is pointer-to-void, static_cast is more accurate (and safer). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
There is no need to use reinterpret_cast when the source or destination type is pointer-to-void, static_cast is more accurate (and safer). Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Given that message is of type pointer-to-void, a static_cast is enough. Signed-off-by:Jean-Baptiste Kempf <jb@videolan.org> -
Rémi Denis-Courmont authored -
Rémi Denis-Courmont authored
This is rarely used (only once in the whole code base).
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
