public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
From: Di Chen <dichen@redhat.com>
To: Aaron Merey <amerey@redhat.com>
Cc: systemtap@sourceware.org
Subject: Re: [PATCH] PR29997: Fix the symbol aliases search failure when symbol version is missing
Date: Wed, 20 Dec 2023 10:56:38 +0800	[thread overview]
Message-ID: <CAN-Pu7QsExY_i3f+mdFZn5T3YWxiPWaqgujfzOPMR8C2mmZzNw@mail.gmail.com> (raw)
In-Reply-To: <CAJDtP-QXdtpdOxoi7_p7iF=OryE7+Wq-YBP-qChgaz8m0nti1w@mail.gmail.com>


[-- Attachment #1.1: Type: text/plain, Size: 8683 bytes --]

Hey Aaron,

I have added the wildcard handling for the PR29997.

From 0c95ce19272d6e13abf6f41c4ae29b6aa925ca20 Mon Sep 17 00:00:00 2001
From: Di Chen <dichen@redhat.com>
Date: Sun, 5 Nov 2023 11:23:50 +0800
Subject: [PATCH] PR29997: Fix the symbol aliases search failure when symbol
 version is missing

After calling module_info::update_symtab, function aliases will be
populated. Then the updated symtab will be used for symbol searching.

For the _IO_new_fopen familty with the aliases:

  $ eu-readelf -s /lib64/libc.so.6  | grep 0000000000077440
    247: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64@
@GLIBC_2.2.5
   1014: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@
@GLIBC_2.2.5
   1028: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 _IO_fopen@
@GLIBC_2.2.5
   1556: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16 _IO_fopen64
   3471: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16 __new_fopen
   4765: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16
_IO_new_fopen
   5110: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64
   7198: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@
@GLIBC_2.2.5
   7433: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 _IO_fopen@
@GLIBC_2.2.5

  a) fopen@@GLIBC_2.2.5 exists in the updated symtab
  b) fopen does not exist in the updated symtab

This PR is to add a version info padding when symbol cannot be found in
the updated symtab.

Signed-off-by: Di Chen <dichen@redhat.com>
---
 dwflpp.cxx | 44 +++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/dwflpp.cxx b/dwflpp.cxx
index 9fccca0a9..044b0ca97 100644
--- a/dwflpp.cxx
+++ b/dwflpp.cxx
@@ -1055,6 +1055,15 @@ dwflpp::iterate_over_functions<void>(int
(*callback)(Dwarf_Die*, void*),
     }

   auto range = v->equal_range(function);
+  // version padding if the symbol is not found
+  if (range.first == range.second)
+    {
+    std::string function_with_ver = function + "@";
+    for (auto it = v->begin(); it != v->end(); ++it)
+      if (it->first.find(function_with_ver) == 0)
+        function_with_ver = it->first;
+    range = v->equal_range(function_with_ver);
+    }
   if (range.first != range.second)
     {
       for (auto it = range.first; it != range.second; ++it)
@@ -1098,7 +1107,10 @@ dwflpp::iterate_over_functions<void>(int
(*callback)(Dwarf_Die*, void*),
           if (pending_interrupts) return DWARF_CB_ABORT;
           const string& func_name = it->first;
           Dwarf_Die& die = it->second;
-          if (function_name_matches_pattern (func_name, function))
+
+          // version padding if the pattern is not matched
+          if ((function_name_matches_pattern (func_name, function)) ||
+              (function_name_matches_pattern (func_name, function + "@*")))
             {
               if (sess.verbose > 4)
                 clog << _F("function cache %s:%s match %s vs %s",
module_name.c_str(),
@@ -1141,6 +1153,15 @@ dwflpp::iterate_single_function<void>(int
(*callback)(Dwarf_Die*, void*),
     }

   auto range = v->equal_range(function);
+  // version padding if the symbol is not found
+  if (range.first == range.second)
+    {
+    std::string function_with_ver = function + "@";
+    for (auto it = v->begin(); it != v->end(); ++it)
+      if (it->first.find(function_with_ver) == 0)
+        function_with_ver = it->first;
+    range = v->equal_range(function_with_ver);
+    }
   if (range.first != range.second)
     {
       for (auto it = range.first; it != range.second; ++it)
@@ -1158,6 +1179,27 @@ dwflpp::iterate_single_function<void>(int
(*callback)(Dwarf_Die*, void*),
           if (rc != DWARF_CB_OK) break;
         }
     }
+  else if (name_has_wildcard (function))
+    {
+      for (auto it = v->begin(); it != v->end(); ++it)
+        {
+          if (pending_interrupts) return DWARF_CB_ABORT;
+          const string& func_name = it->first;
+          Dwarf_Die& die = it->second;
+
+          // version padding if the pattern is not matched
+          if ((function_name_matches_pattern (func_name, function)) ||
+              (function_name_matches_pattern (func_name, function + "@*")))
+            {
+              if (sess.verbose > 4)
+                clog << _F("function cache %s:%s match %s vs %s",
module_name.c_str(),
+                           cu_name().c_str(), func_name.c_str(),
function.c_str()) << endl;
+
+              rc = (*callback)(& die, data);
+              if (rc != DWARF_CB_OK) break;
+            }
+        }
+    }

   // undo the focus_on_cu
   this->cu = NULL;
-- 
2.41.0


On Tue, Nov 21, 2023 at 12:05 AM Aaron Merey <amerey@redhat.com> wrote:

> Hi Di,
>
> Thanks for working on this.
>
> On Sun, Nov 5, 2023 at 12:00 AM Di Chen via Systemtap
> <systemtap@sourceware.org> wrote:
> >
> > From 9273dc95658b109f048e4ec9b0fcde96e34f3419 Mon Sep 17 00:00:00 2001
> > From: Di Chen <dichen@redhat.com>
> > Date: Sun, 5 Nov 2023 11:23:50 +0800
> > Subject: [PATCH] PR29997: Fix the symbol aliases search failure when
> symbol
> >  version is missing
> >
> > After calling module_info::update_symtab, function aliases will be
> > populated. Then the updated symtab will be used for symbol searching.
> >
> > For the _IO_new_fopen familty with the aliases:
> >
> >   $ eu-readelf -s /lib64/libc.so.6  | grep 0000000000077440
> >     247: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64@
> > @GLIBC_2.2.5
> >    1014: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@
> > @GLIBC_2.2.5
> >    1028: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16
> _IO_fopen@
> > @GLIBC_2.2.5
> >    1556: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16
> _IO_fopen64
> >    3471: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16
> __new_fopen
> >    4765: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16
> > _IO_new_fopen
> >    5110: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64
> >    7198: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@
> > @GLIBC_2.2.5
> >    7433: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16
> _IO_fopen@
> > @GLIBC_2.2.5
> >
> >   a) fopen@@GLIBC_2.2.5 exists in the updated symtab
> >   b) fopen does not exist in the updated symtab
> >
> > This PR is to add a version info padding when symbol cannot be found in
> > the updated symtab.
> >
> > Signed-off-by: Di Chen <dichen@redhat.com>
> > ---
> >  dwflpp.cxx | 18 ++++++++++++++++++
> >  1 file changed, 18 insertions(+)
> >
> > diff --git a/dwflpp.cxx b/dwflpp.cxx
> > index a4f66440f..22e1479c9 100644
> > --- a/dwflpp.cxx
> > +++ b/dwflpp.cxx
> > @@ -1052,6 +1052,15 @@ dwflpp::iterate_over_functions<void>(int
> > (*callback)(Dwarf_Die*, void*),
> >      }
> >
> >    auto range = v->equal_range(function);
> > +  // version info padding if the symbol is not found
> > +  if (range.first == range.second)
> > +    {
> > +    std::string function_with_ver = function + "@";
> > +    for (auto it = v->begin(); it != v->end(); ++it)
> > +      if (it->first.find(function_with_ver) == 0)
> > +        function_with_ver = it->first;
> > +    range = v->equal_range(function_with_ver);
> > +    }
>
> This fixes the bug for function names without wildcards.  However
> the bug can still be reproduced using function names with wildcards.
> For example the following command incorrectly returns no results:
>
>   $ stap -L 'process("/lib64/libc.so.6").function("fo*en")'
>
> We'll need additional '@' handling in the section of
> dwflpp::iterate_over_functions that matches wildcards.
>
> >    if (range.first != range.second)
> >      {
> >        for (auto it = range.first; it != range.second; ++it)
> > @@ -1138,6 +1147,15 @@ dwflpp::iterate_single_function<void>(int
> > (*callback)(Dwarf_Die*, void*),
> >      }
> >
> >    auto range = v->equal_range(function);
> > +  // version info padding if the symbol is not found
> > +  if (range.first == range.second)
> > +    {
> > +    std::string function_with_ver = function + "@";
> > +    for (auto it = v->begin(); it != v->end(); ++it)
> > +      if (it->first.find(function_with_ver) == 0)
> > +        function_with_ver = it->first;
> > +    range = v->equal_range(function_with_ver);
> > +    }
> >    if (range.first != range.second)
> >      {
> >        for (auto it = range.first; it != range.second; ++it)
> > --
> > 2.41.0
>
> Aaron
>
>

[-- Attachment #2: 0001-PR29997-Fix-the-symbol-aliases-search-failure-when-s.patch --]
[-- Type: text/x-patch, Size: 4585 bytes --]

From 0c95ce19272d6e13abf6f41c4ae29b6aa925ca20 Mon Sep 17 00:00:00 2001
From: Di Chen <dichen@redhat.com>
Date: Sun, 5 Nov 2023 11:23:50 +0800
Subject: [PATCH] PR29997: Fix the symbol aliases search failure when symbol
 version is missing

After calling module_info::update_symtab, function aliases will be
populated. Then the updated symtab will be used for symbol searching.

For the _IO_new_fopen familty with the aliases:

  $ eu-readelf -s /lib64/libc.so.6  | grep 0000000000077440
    247: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64@@GLIBC_2.2.5
   1014: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@@GLIBC_2.2.5
   1028: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 _IO_fopen@@GLIBC_2.2.5
   1556: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16 _IO_fopen64
   3471: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16 __new_fopen
   4765: 0000000000077440     14 FUNC    LOCAL  DEFAULT       16 _IO_new_fopen
   5110: 0000000000077440     14 FUNC    WEAK   DEFAULT       16 fopen64
   7198: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 fopen@@GLIBC_2.2.5
   7433: 0000000000077440     14 FUNC    GLOBAL DEFAULT       16 _IO_fopen@@GLIBC_2.2.5

  a) fopen@@GLIBC_2.2.5 exists in the updated symtab
  b) fopen does not exist in the updated symtab

This PR is to add a version info padding when symbol cannot be found in
the updated symtab.

Signed-off-by: Di Chen <dichen@redhat.com>
---
 dwflpp.cxx | 44 +++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/dwflpp.cxx b/dwflpp.cxx
index 9fccca0a9..044b0ca97 100644
--- a/dwflpp.cxx
+++ b/dwflpp.cxx
@@ -1055,6 +1055,15 @@ dwflpp::iterate_over_functions<void>(int (*callback)(Dwarf_Die*, void*),
     }
 
   auto range = v->equal_range(function);
+  // version padding if the symbol is not found
+  if (range.first == range.second)
+    {
+    std::string function_with_ver = function + "@";
+    for (auto it = v->begin(); it != v->end(); ++it)
+      if (it->first.find(function_with_ver) == 0)
+        function_with_ver = it->first;
+    range = v->equal_range(function_with_ver);
+    }
   if (range.first != range.second)
     {
       for (auto it = range.first; it != range.second; ++it)
@@ -1098,7 +1107,10 @@ dwflpp::iterate_over_functions<void>(int (*callback)(Dwarf_Die*, void*),
           if (pending_interrupts) return DWARF_CB_ABORT;
           const string& func_name = it->first;
           Dwarf_Die& die = it->second;
-          if (function_name_matches_pattern (func_name, function))
+
+          // version padding if the pattern is not matched
+          if ((function_name_matches_pattern (func_name, function)) ||
+              (function_name_matches_pattern (func_name, function + "@*")))
             {
               if (sess.verbose > 4)
                 clog << _F("function cache %s:%s match %s vs %s", module_name.c_str(),
@@ -1141,6 +1153,15 @@ dwflpp::iterate_single_function<void>(int (*callback)(Dwarf_Die*, void*),
     }
 
   auto range = v->equal_range(function);
+  // version padding if the symbol is not found
+  if (range.first == range.second)
+    {
+    std::string function_with_ver = function + "@";
+    for (auto it = v->begin(); it != v->end(); ++it)
+      if (it->first.find(function_with_ver) == 0)
+        function_with_ver = it->first;
+    range = v->equal_range(function_with_ver);
+    }
   if (range.first != range.second)
     {
       for (auto it = range.first; it != range.second; ++it)
@@ -1158,6 +1179,27 @@ dwflpp::iterate_single_function<void>(int (*callback)(Dwarf_Die*, void*),
           if (rc != DWARF_CB_OK) break;
         }
     }
+  else if (name_has_wildcard (function))
+    {
+      for (auto it = v->begin(); it != v->end(); ++it)
+        {
+          if (pending_interrupts) return DWARF_CB_ABORT;
+          const string& func_name = it->first;
+          Dwarf_Die& die = it->second;
+
+          // version padding if the pattern is not matched
+          if ((function_name_matches_pattern (func_name, function)) ||
+              (function_name_matches_pattern (func_name, function + "@*")))
+            {
+              if (sess.verbose > 4)
+                clog << _F("function cache %s:%s match %s vs %s", module_name.c_str(),
+                           cu_name().c_str(), func_name.c_str(), function.c_str()) << endl;
+
+              rc = (*callback)(& die, data);
+              if (rc != DWARF_CB_OK) break;
+            }
+        }
+    }
 
   // undo the focus_on_cu
   this->cu = NULL;
-- 
2.41.0


  reply	other threads:[~2023-12-20  2:56 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-11-05  3:59 Di Chen
2023-11-20 16:05 ` Aaron Merey
2023-12-20  2:56   ` Di Chen [this message]
2024-01-18  0:15     ` Aaron Merey
2024-01-31  2:45       ` Di Chen
2024-01-31 22:38         ` Aaron Merey

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=CAN-Pu7QsExY_i3f+mdFZn5T3YWxiPWaqgujfzOPMR8C2mmZzNw@mail.gmail.com \
    --to=dichen@redhat.com \
    --cc=amerey@redhat.com \
    --cc=systemtap@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).