Update to recent upstream master #1
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Given the throttle plugin is collecting timing statistics for I/O anyway, have it begin to report these statistics out through the g-stream. This allows the monitoring programs to determine the overall time spent in the I/O subsystem.
Need to check if bit for noDread is enabled in dOpts for the check to work correctly. Signed-off-by: Guilherme Amadio <[email protected]>
Setting this on all paths causes directories to appear as files when using oss.rsscmd option. Signed-off-by: Guilherme Amadio <[email protected]>
In Alma 9.3 the default curl version is 7.76.1 and according to the documentation [1], starting with 7.62.0 the default HTTP protocol used is HTTP 2 TLS. Since XRootD supports only HTTP 1.1, this leads to the following error whenever a HTTP TPC transfer is attempted: "HTTP library failure: Stream error in the HTTP/2 framing layer" This patch will force curl to use HTTP 1.1 without even attempting HTTP 2. This is a workaround for the currently available curl version since normaly curl should be able to gracefully downgrade from HTTP 2 to 1.1 by itself. [1] https://curl.se/libcurl/c/CURLOPT_HTTP_VERSION.html
The function InitTLS() can be called either from the constructor of XrdCl::Tls(), or from XRootDTransport::InitProtocolReq(), while the protocol initialization is happening. Since these calls can happen concurrently, for example, when called as a socket callback during the initial connection setup, there can be a crash if two threads try to assign to the unique_ptr at the same time. Fixes: #2220
Caused by .first in the second statement. Since we are touching these lines, use map.emplace() to simplify the code as well.
Required due to GitHub Actions migration to node 20: https://github.blog/changelog/2023-09-22-github-actions-transitioning-from-node-16-to-node-20 Note: CentOS 7 checkout fails with v4, so must use v1 (will be removed at EOL later).
In some cases, a token may be mapped to a known user - but we only want to utilize the mapping if there's scope-based authorization in place. This can now be controlled as part of the scitokens.cfg. In the issuer section, one can add: ``` authorization_strategy = strategy1 strategy2 strategy3 ``` Where possible strategies are `capability` (if authorized based on scope, pass through the username mapping), `group` (if authorized based on scope, pass through the group information), and `mapping`, (always pass through the explicit mapping, even if there's no scope present).
std::result_of is deprecated in C++17 and removed in C++20.
Required due to GitHub Actions migration to node 20: https://github.blog/changelog/2023-09-22-github-actions-transitioning-from-node-16-to-node-20
The comparison path[rule.second.size()] == '/' for ensuring path is a subdirectory of the rule's path does not work for '/'. For example if the rule's path is '/' and requested path is '/tmp', then the comparison above checks path[rule.second.size()] == path[1] == '/', but actually, path[1] == 't'.
Between the call to stat and the open, the file size may change. First open the file, then use fstat with the file descriptor to ensure consistency.
First open then fstat, instead of stat then fopen, to guarantee that the information from stat and open are consistent with each other. Afterwards, convert file descriptor to FILE* with fdopen. Ensure also that the file is closed should an error occur.
Make sure that the information is consistent by first opening and subsequently using the same file descriptor for all operations. Since we needed to add an include as well, make sure to include our own headers first, then system headers.
The defaultID is created with defaultID = new XrdSecsssEnt(...); so it cannot be deallocated with free(defaultID). Since the destructor is private, we cannot call delete defaultID either. We must call the specific function for this, which is defaultID->Delete(), which in turn will call delete instead of free to deallocate the memory.
Using pointers in expressions that may cause overflow leads to undefined behavior. See also https://lwn.net/Articles/278137/
The variable jkey is already checked to be non-null a few lines above.
In line 707, there is already an if(!ifList || !(*ifList)) which makes the check in line 732 redundant.
The multiplication of int * int might be bigger than an int can hold, but still fits in a size_t. Convert before multiplying to ensure no overflow.
The multiplication is numq * xfrovhd, which if carried with ints can overflow before being converted to unsigned long long. Convert one of the operands to unsigned long long to avoid this.
From the man page for snprintf:
The functions snprintf() and vsnprintf() do not write more than size
bytes (including the terminating null byte ('\0')). If the output was
truncated due to this limit, then the return value is the number of
characters (excluding the terminating null byte) *which would have been
written to the final string if enough space had been available*. Thus,
a return value of size or more means that the output was truncated.
We therefore need to check if n > buffer size, and act accordingly.
The action to upload-artifact@v3 no longer works due to its reliance on node16, which has just hit EOL and been removed. Since upload-artifact@v4 also doesn't work on CentOS 7, the only option to keep builds running is to not upload results. https://github.blog/changelog/2024-04-16-deprecation-notice-v3-of-the-artifact-actions/
When moving to actions/checkout@v4, v1 was kept on Alpine Linux by mistake. The previous commit moved CentOS 7 properly to v1, and this commit fixes the problem in the Alpine Linux build. Fixes: 9096834
This reverts commit 50209b3. The fix is incorrect, as struct XrdSecsssRR_Data actually derives from XrdSecsssRR_DataHdr, which introduces some members in between such that prData.Data is not actually at the beginning of the struct, but at some non-zero offset (i.e. different address than &prData).
This was an oversight when adding the option in commit below. Fixes: a5cfef7
If a test causes a server to crash, a pid file with the pid of the crashed server will be left behind. The teardown phase would then fail due to a non-zero return of the ps command when assigning the pid to $PID. This changes the test to skip killing a process that has already exited without failing.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.