Re: [PATCH v3] Add debuginfod support to GDB

[Tom Tromey <tom@tromey.com>]

>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:

Simon> On 2020-02-20 1:02 p.m., Simon Marchi wrote:
>> I think we still need something just after the debuginfod_find_debuginfo
>> and debuginfod_find_source calls to throw the "quit" exception, if the quit
>> flag is set at this point.  Without that, the debuginfod_source_query and
>> debuginfod_debuginfo_query will return a failure.  The calling code could
>> therefore conclude that debug info is not available (and cache that value
>> somehwere), for a file for which debug info is actually available.  When
>> the execution reaches the event loop and we process the quit flag, it
>> would be too late, the damage will have been done.

Simon> Following a discussion on IRC, I changed my mind, the simplest solution
Simon> for now would be: when the user cancels a download, GDB will continue
Simon> as if debug info was not available for that objfile.  However, I would suggest
Simon> printing a message saying that the download of debug info for XYZ was cancelled,
Simon> so the user is not surprised to not have its debug info.

Wouldn't this imply also clearing (or more precisely - not preserving)
the quit flag in the interrupt case?

It seems to me that as long as the interrupt is processed by the
library, then we wouldn't want to set the quit flag again.  Otherwise
that will look like a double interrupt.

Tom