So far idr support was available for only radix-tree based idrs.
Thus radix-tree tests were implicity covering idr test as well.
Now as we are supporting non radix-tree based idrs as well, so
add explicit test cases for idr testing.
The test are applicable for both new (i.e radix-tree based) and
old implementation of idrs.
Signed-off-by: Imran Khan <imran.f.khan@oracle.com>
Prior to kernel v4.11, idr was not using radix tree as its backend.
So current idr helper(s) only work for kernel v4.11+.
Enable idr helpers(s) to work with non-radix tree based idr, so that
the helpers can be used with older kernels as well.
Thanks to Omar for optimizing the idr_for_each helper.
Signed-off-by: Imran Khan <imran.f.khan@oracle.com>
The CI keeps has been sporadically hitting errors like:
downloading kernel release 6.0.19-vmtest17.1default to build/vmtest/x86_64/kernel-6.0.19-vmtest17.1default from https://api.github.com/repos/osandov/drgn/releases/assets/91622017
/*stdin*\ : Read error (39) : premature end
tar: Unexpected EOF in archive
tar: Unexpected EOF in archive
tar: Error is not recoverable: exiting now
The only way I can see this happening is if the download is being
truncated, so add a check (inspired by urllib.request.urlretrieve()).
Signed-off-by: Omar Sandoval <osandov@osandov.com>
The parameters of the read_file function in the example are swapped.
This matches the parameters order used by py_memory_read_fn.
Fixes: c0bc72b0ea ("libdrgn: use splay tree for memory reader")
Signed-off-by: Timothée Cocault <timothee.cocault@gmail.com>
Looking up objects in namespaces is already well-supported by `drgn`.
These changes bring the same to functionality type lookup, so that
`prog.type('struct A::B::C::MyType')` works in an analogous fashion to
`prog['A::B::C::MyVar']`.
Signed-off-by: Kevin Svetlitski <svetlitski@meta.com>
Object attributes which are not changed after the object's creation
have been marked with the `Final` type attribute. The primary utility
of this change is that some Python type checkers, [such as
Pyre](https://pyre-check.org/docs/errors/#optional-attributes), rely on
the presence of the `Final` attribute to determine whether `Optional`
values have been validated as being present before their contents are
accessed.
Signed-off-by: Kevin Svetlitski <svetlitski@meta.com>
drgn is able to lookup some special metadata for the Linux kernel,
sometimes even without debuginfo. For users who may expect that
Program.object() will only return objects corresponding to variables,
these metadata are unexpected and can be quite useful. Regardless,
they're currently undocumented. Add documentation under Advanced Usage
for these, and reference it in the Program.object() docstring.
Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Lots of core tools expect Unix sockets to work. Stephen Brennan recently
hit issue where installing Ksplice on Ubuntu replaced modprobe with a
version that uses Unix sockets, which caused the kernel to call modprobe
to load the Unix socket module, recursively until the kernel's modprobe
limit of 50 was hit.
Enable Unix sockets to avoid other surprises like this. (We'll probably
eventually have drgn helpers for Unix sockets, too.)
Signed-off-by: Omar Sandoval <osandov@osandov.com>
It's allowable for task->files to be NULL. Handle this case gracefully
by returning without yielding any files.
Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Reported-by: Wengang Wang <wen.gang.wang@oracle.com>
I've been posting these internally at Meta for previous releases, but I
think others would find them useful, too. In the future, I'll write them
up before cutting the release.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
manylinux2010 has been EOL since August 2022; see pypa/manylinux#1281.
In particular, the manylinux2010 image wasn't updated with Python 3.11.
Update scripts/build_dist.sh to use manylinux2014 and drop some CentOS 6
workarounds from scripts/build_manylinux_in_docker.sh.
If someone really wants manylinux2010, we can probably build both, but
none of the popular packages I checked (NumPy, SciPy, cryptography,
charset-normalizer) do this.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
Commit c48cddbdb0 ("libdrgn: ppc64: fix stack unwinding since Linux
v5.11 and before v4.20") is the only change necessary for Linux v6.2.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
get_registers_from_frame_pointer() has a sanity check that the unwound
frame pointer must be greater than the current frame pointer. This is
generally true if the entire program is using frame pointers, but not
necessarily otherwise. In particular, if the program is a Linux kernel
configured with ORC, most of the time, rbp is a general purpose
register; it is only used as a frame pointer in special cases without
unwinder information like BPF programs. Those cases are exactly when we
want the frame pointer unwinder, but depending on what the caller was
using rbp for, the frame pointer unwinder might bail prematurely.
Let's remove the sanity check. In the worst case, this could lead us off
into the weeds chasing pointers, but the iteration limit in
drgn_get_stack_trace() prevents that from being dangerous.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
ppc64's crash_setup_regs() calls ppc_save_regs(), which isn't exported.
So, we need to provide our own implementation. At this point, we might
as well just copy the implementations of crash_setup_regs() for x86-64
and AArch64 and stop trying to use crash_setup_regs().
Signed-off-by: Omar Sandoval <osandov@osandov.com>
When the architecture is missing virtual address translation or the
kernel is configured with SLOB, print_annotated_stack() will not
identify slab objects.
Fixes: 05041423c7 ("drgn.helpers.common.stack: add print_annotated_stack helper function")
Signed-off-by: Omar Sandoval <osandov@osandov.com>
identify_address() is supposed to be best-effort, but if it is called on
an architecture without virtual address translation support, it will
raise a NotImplementedError. Let's catch and ignore it.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
It's useful to check in on a long run without needing to scroll back and
check whether each kernel version passed or failed.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
linux_kernel_get_initial_registers_ppc64() depends on the size of struct
pt_regs, but this has changed multiple times, in:
- Linux kernel commit 4c2de74cc869 ("powerpc/64: Interrupts save PPR on stack rather than
thread_struct") (in v4.20)
- Linux kernel commit 66f93c5a02d5 ("powerpc/64: Fix kernel stack
16-byte alignment") (in v4.20)
- Linux kernel commit 8e560921b58c ("powerpc/book3s64/pkeys:
Store/restore userspace AMR/IAMR correctly on entry and exit from
kernel") (in v5.11)
It also depends on the overhead stored before struct pt_regs on the
stack, which changed in Linux kernel commit cd52414d5a6c ("powerpc/64:
ELFv2 use minimal stack frames in int and switch frame sizes") (in
v6.2).
We can handle all of these cases by reading the previous r1 from memory
instead of computing it from a hard-coded size and finding the struct
pt_regs based on that r1 and the actual size of struct pt_regs.
Reported in #232.
Reported-by: Sourabh Jain <jainsourabh679@gmail.com>
Signed-off-by: Omar Sandoval <osandov@osandov.com>
I changed the default workflow permissions to be restrictive, but I
forgot that the vmtest build needs to upload assets. Add write
permissions for the contents scope, which seems to cover asset uploads.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
The vmtest build is silently failing to upload kernel builds. We're not
checking for HTTP errors, so let's see if that provides some
information.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
This should fix the "HTTP Error 403: rate limit exceeded" errors we've
been seeing. (I don't know if GitHub Actions used to do this
automatically or they started being stricter about rate limits.)
Signed-off-by: Omar Sandoval <osandov@osandov.com>
We're getting a bunch of test failures with "HTTP Error 403: rate limit
exceeded" from the GitHub API. The API apparently includes additional
information in the response, so let's print that so that we can figure
out which limit we're hitting.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
The Linux kernel apparently enables -Wundef, so although #if is correct,
it results in a warning. Use #ifdef instead.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
The only relocation type I saw in Debian's kernel module debug info was
R_390_32. R_390_8, R_390_16, R_390_64, R_390_PC16, R_390_PC32, and
R_390_PC64 are trivial to support, as well. The Linux kernel supports
many more, but hopefully they won't show up for debug info.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
I've had lots of reports that turned out to be due to missing debug info
(e.g., #161, #175, #231, and more offline). A big part of the problem is
probably that the warning gets lost among the startup information. Let's
move the warning to be the last thing we print and also color it.
After the debug info discovery rework, we can print more specific
advice, but this is better than nothing.
Closes#41.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
Konrad Borowski pointed out that add_to_possibly_null_pointer() relies
on GCC-specific behavior:
https://fosstodon.org/@xfix/109542070338182493. CONTRIBUTING.rst
mentions that we assume that casting between pointers and integers does
not change the bit representation, but we might as well document it
here, too.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
The current test case only checks that getting a stack trace succeeds.
By having the test kernel module create a struct pt_regs, we can
actually test that we get a reasonable stack trace.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
for_each_page() can return offline pages that don't actually exist.
Fixing this is difficult (see #228), but working around it by catching
FaultError is easy. Document the recommended usage.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
CONFIG_SLOB was renamed to CONFIG_SLOB_DEPRECATED during the Linux 6.2
merge window. Let's prepare for that now before rc1 is released.
Signed-off-by: Omar Sandoval <osandov@osandov.com>
Currently, looking up a type with template arguments results in an
"invalid character" syntax error on the "<" character. The DWARF index
includes template arguments in indexed names, so we need to do lookups
including the template arguments. Full support for this would require
parsing the template argument list syntax and normalizing it or looking
it up as an AST in some way. For now, it's at least an improvement to
pass the user's string verbatim. To do so, kludge it by adding a token
containing everything from "<" to the matching ">" to the C++ lexer and
appending that to the identifier.
Co-authored-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Kevin Svetlitski <svetlitski@meta.com>
Commit 89eb868e95 ("helpers: make find_task() work on recent kernels")
made radix_tree_lookup() and radix_tree_for_each() work for basic
XArrays. However, it doesn't handle a couple of more advanced features:
multi-index entries (which old radix trees actually also supported) and
zero entries. It has also been really confusing to explain to people
unfamiliar with the radix tree -> XArray transition that they should use
helpers named radix_tree for a structure named xarray.
So, let's finally add xa_load(), xa_for_each(), and some additional
auxiliary helpers. The non-recursive xa_for_each() implementation is
based on Kevin Svetlitski's C implementation from commit 2b47583c73
("Rewrite linux helper iterators in C"). radix_tree_lookup() and
radix_tree_for_each() share the implementation with xa_load() and
xa_for_each(), respectively, so they are mostly interchangeable.
Fixes: #61
Signed-off-by: Omar Sandoval <osandov@osandov.com>
