From: watashiwaher <watashiwaher@gmail.com>
To: gdb-patches@sourceware.org
Subject: [PATCH] Fixing get_builder() function in dwarf2/read.c
Date: Mon, 15 Jun 2020 01:49:21 +0900 [thread overview]
Message-ID: <CAFHDqJOXn6P_U2gP4FvSJa9NB3kOGQj5P4dan84jPxJJpX9LFg@mail.gmail.com> (raw)
Explanation
I found that some binaries with debug symbols halt gdb. And the problem was
in the get_builder() function inside dwarf2/read.c file. The function
recursively called itself until infinity. So I tried to fix this issue by
eliminating recursion and adding the check for cycle detection, and
returning nullptr if cycle detected.
ChangeLog
2020-06-15 Slava Mitin <watashiwaher@gmail.com>
* dwarf2/read.c (get_builder): Added check for cycles
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index e3073fe43c..e8c6d3bec3 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -594,15 +594,51 @@ struct dwarf2_cu
/* Get the buildsym_compunit for this CU. */
buildsym_compunit *get_builder ()
{
- /* If this CU has a builder associated with it, use that. */
- if (m_builder != nullptr)
- return m_builder.get ();
+ /* Number of iterations before the updating of
+ the recursion_check_cu pointer */
+ const int RECURSION_CHECK_UPDATE_INTERVAL = 2;
- /* Otherwise, search ancestors for a valid builder. */
- if (ancestor != nullptr)
- return ancestor->get_builder ();
+ /* Replacing recursion calls of get_builder
+ to the loop */
+ dwarf2_cu *current_cu = this;
- return nullptr;
+ /* Sometimes CUs can include each other recuresively,
+ so we need to stop iterating some day, if they do */
+ dwarf2_cu *recursion_check_cu = nullptr;
+
+ /* Counter of intervals */
+ int recursion_check_counter = 1;
+
+ while (true) {
+ /* If this CU has a builder associated with it, use that. */
+ if (current_cu->m_builder != nullptr) {
+ return current_cu->m_builder.get();
+ }
+
+ /* Otherwise, search ancestors for a valid builder. */
+ if (current_cu->ancestor != nullptr) {
+ current_cu = current_cu->ancestor;
+ } else {
+ return nullptr;
+ }
+
+ /* If CU was seen already, return. */
+ if (recursion_check_cu == current_cu) {
+ return nullptr;
+ }
+
+ /* Update the variable used to check the recursion */
+ if (recursion_check_counter == RECURSION_CHECK_UPDATE_INTERVAL) {
+ if (recursion_check_cu == nullptr) {
+ recursion_check_cu = this;
+ } else {
+ recursion_check_cu = recursion_check_cu->ancestor;
+ }
+ recursion_check_counter = 1;
+ } else {
+ recursion_check_counter++;
+ }
+ }
}
};
next reply other threads:[~2020-06-14 16:49 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-14 16:49 watashiwaher [this message]
2020-06-14 16:58 ` Simon Marchi
2020-06-14 17:07 ` watashiwaher
2020-06-15 14:58 ` Simon Marchi
2020-06-15 15:06 ` watashiwaher
2020-06-15 15:13 ` watashiwaher
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAFHDqJOXn6P_U2gP4FvSJa9NB3kOGQj5P4dan84jPxJJpX9LFg@mail.gmail.com \
--to=watashiwaher@gmail.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).