Re: [PATCH v3 bpf-next 1/4] libbpf: support function name-based attach uprobes

[Alan Maguire <alan.maguire@oracle.com>]

On Fri, 4 Feb 2022, Andrii Nakryiko wrote:

> On Mon, Jan 31, 2022 at 8:13 AM Alan Maguire <alan.maguire@oracle.com> wrote:
> >
> > kprobe attach is name-based, using lookups of kallsyms to translate
> > a function name to an address.  Currently uprobe attach is done
> > via an offset value as described in [1].  Extend uprobe opts
> > for attach to include a function name which can then be converted
> > into a uprobe-friendly offset.  The calcualation is done in
> > several steps:
> >
> > 1. First, determine the symbol address using libelf; this gives us
> >    the offset as reported by objdump; then, in the case of local
> >    functions
> > 2. If the function is a shared library function - and the binary
> >    provided is a shared library - no further work is required;
> >    the address found is the required address
> > 3. If the function is a shared library function in a program
> >    (as opposed to a shared library), the Procedure Linking Table
> >    (PLT) table address is found (it is indexed via the dynamic
> >    symbol table index).  This allows us to instrument a call
> >    to a shared library function locally in the calling binary,
> >    reducing overhead versus having a breakpoint in global lib.
> > 4. Finally, if the function is local, subtract the base address
> >    associated with the object, retrieved from ELF program headers.
> >
> > The resultant value is then added to the func_offset value passed
> > in to specify the uprobe attach address.  So specifying a func_offset
> > of 0 along with a function name "printf" will attach to printf entry.
> >
> > The modes of operation supported are then
> >
> > 1. to attach to a local function in a binary; function "foo1" in
> >    "/usr/bin/foo"
> > 2. to attach to a shared library function in a binary;
> >    function "malloc" in "/usr/bin/foo"
> > 3. to attach to a shared library function in a shared library -
> >    function "malloc" in libc.
> >
> > [1] https://www.kernel.org/doc/html/latest/trace/uprobetracer.html
> >
> > Signed-off-by: Alan Maguire <alan.maguire@oracle.com>
> > ---
> 
> This looks great and very clean. I left a few nits, but otherwise it
> looks ready (still need to go through the rest of the patches)
> 
> >  tools/lib/bpf/libbpf.c | 250 +++++++++++++++++++++++++++++++++++++++++++++++++
> >  tools/lib/bpf/libbpf.h |  10 +-
> >  2 files changed, 259 insertions(+), 1 deletion(-)
> >
>

<snip>
 
> if both the symbol name and requested function name have @ in them,
> what should be the comparison rule? Shouldn't it be an exact match
> including '@@' and part after it?
>

In this case, we might want to support matching on malloc@GLIBC and
malloc@GLIBC_2.3.4; in other words letting the caller decide how
specific they want to be makes sense I think.  So the caller dictates
the matching length via the argument they provide - with the proviso that
if it's just a function name without a "@LIBRARY" suffix it must match 
fully. The problem with the version numbers associated with functions is 
they're the versions from the mapfiles, so the same library version has 
malloc@GLIBC_2.2.5, epoll_ctl@GLIBC_2.3.2 etc.

Thanks for the review! I'm working on incorporating all of these changes
into v4 now.

Alan