Disallow more unsafe string->path conversions allowed by path append operators #24470

ryanofsky · 2022-03-03T20:07:34Z

Add more fs::path operator/ and operator+ overloads to prevent unsafe string->path conversions on Windows that would cause strings to be decoded according to the current Windows locale & code page instead of the correct string encoding.

Update application code to deal with loss of implicit string->path conversions by calling fs::u8path or fs::PathFromString explicitly, or by just changing variable types from std::string to fs::path to avoid conversions altogether, or make them happen earlier.

In all cases, there's no change in behavior either (1) because strings only contained ASCII characters and would be decoded the same regardless of what encoding was used, or (2) because of the 1:1 mapping between paths and strings using the PathToString and PathFromString functions.

Motivation for this PR was just that I was experimenting with #24469 and noticed that operations like fs::path / std::string were allowed, and I thought it would be better not to allow them.

DrahtBot · 2022-03-03T22:34:22Z

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Conflicts

Reviewers, this pull request conflicts with the following ones:

#18904 (Don't call lsn_reset in periodic flush by bvbfan)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

hebasto · 2022-03-05T19:24:26Z

Concept ACK.

laanwj · 2022-04-07T14:24:43Z

Concept ACK

…operators Add more fs::path operator/ and operator+ overloads to prevent unsafe string->path conversions on Windows that would cause strings to be decoded according to the current Windows locale & code page instead of the correct string encoding. Update application code to deal with loss of implicit string->path conversions by calling fs::u8path or fs::PathFromString explicitly, or by just changing variable types from std::string to fs::path to avoid conversions altoghther, or make them happen earlier. In all cases, there's no change in behavior either (1) because strings only contained ASCII characters and would be decoded the same regardless of what encoding was used, or (2) because of the 1:1 mapping between paths and strings using the PathToString and PathFromString functions. Co-authored-by: Hennadii Stepanov <32963518+hebasto@users.noreply.github.com>

ryanofsky · 2022-04-22T02:17:39Z

Rebased 67ca71e -> cb9f6b1 (pr/patha.1 -> pr/patha.2, compare) due to conflict with #23732, and adding fix for windows build error https://cirrus-ci.com/task/4639862308470784

hebasto · 2022-04-22T07:45:09Z

Maybe add

diff --git a/src/qt/guiutil.cpp b/src/qt/guiutil.cpp
index 1e67b1a0d..706cf2f7d 100644
--- a/src/qt/guiutil.cpp
+++ b/src/qt/guiutil.cpp
@@ -508,7 +508,7 @@ fs::path static StartupShortcutPath()
         return GetSpecialFolderPath(CSIDL_STARTUP) / "Bitcoin.lnk";
     if (chain == CBaseChainParams::TESTNET) // Remove this special case when CBaseChainParams::TESTNET = "testnet4"
         return GetSpecialFolderPath(CSIDL_STARTUP) / "Bitcoin (testnet).lnk";
-    return GetSpecialFolderPath(CSIDL_STARTUP) / strprintf("Bitcoin (%s).lnk", chain);
+    return GetSpecialFolderPath(CSIDL_STARTUP) / fs::u8path(strprintf("Bitcoin (%s).lnk", chain));
 }
 
 bool GetStartOnSystemStartup()
diff --git a/src/wallet/bdb.cpp b/src/wallet/bdb.cpp
index 82693eae8..dbc297f45 100644
--- a/src/wallet/bdb.cpp
+++ b/src/wallet/bdb.cpp
@@ -302,7 +302,7 @@ BerkeleyDatabase::~BerkeleyDatabase()
         assert(!m_db);
         size_t erased = env->m_databases.erase(m_filename);
         assert(erased == 1);
-        env->m_fileids.erase(m_filename);
+        env->m_fileids.erase(fs::PathToString(m_filename));
     }
 }

?

It makes CI greener -- https://cirrus-ci.com/build/6528392144093184 :)

hebasto · 2022-04-22T09:03:15Z

src/fs.h

    return p1;
 }
+static inline path operator/(path p1, std::filesystem::path p2)


Why std::filesystem::path for second parameter? Isn't it safer to limit the usage of std::filesystem::path only within the fs::path wrapper?

re: #24470 (comment)

Why std::filesystem::path for second parameter? Isn't it safer to limit the usage of std::filesystem::path only within the fs::path wrapper?

Good point, fs::path is safer and this doesn't seem to be necessary, now dropped

dongcarl · 2022-04-22T21:34:39Z

src/fs.h

+static inline path operator/(path p1, path p2)
 {
-    p1 += std::move(p2);
+    p1 /= std::move(p2);
    return p1;
 }
+static inline path operator/(path p1, std::filesystem::path p2)
+{
+    p1 /= std::move(p2);
+    return p1;
+}
+static inline path operator/(path p1, const char* p2)
+{
+    p1 /= p2;
+    return p1;
+}
+static inline path operator+(path p1, const char* p2)
+{
+    p1 += p2;
+    return p1;
+}
+static inline path operator+(path p1, path::value_type p2)
+{
+    p1 += p2;
+    return p1;
+}


Wondering why there's a difference between the allowed right-hand operands for / and +

re: #24470 (comment)

Wondering why there's a difference between the allowed right-hand operands for / and +

The cases that are listed are just cases that seemed safe and are in used by existing code. (I think it'd be fine to add cases that are safe and aren't in use by existing code, too, but I didn't go looking for these.)

On the specific cases, combining two path objects with / is safe and useful. Combining two path objects with + is safe probably not useful. Appending a literal string is safe to a path is safe (assuming the literal string is ASCII) both with + and /. Adding a native character to a path is safe and semi-useful (used by some tests at least)

ryanofsky

Updated cb9f6b1 -> f64aa9c (pr/patha.2 -> pr/patha.3, compare) adding hebasto's fixes for windows build errors https://cirrus-ci.com/build/6528392144093184

ryanofsky · 2022-04-25T16:54:38Z

src/fs.h

    return p1;
 }
+static inline path operator/(path p1, std::filesystem::path p2)


re: #24470 (comment)

Why std::filesystem::path for second parameter? Isn't it safer to limit the usage of std::filesystem::path only within the fs::path wrapper?

Good point, fs::path is safer and this doesn't seem to be necessary, now dropped

ryanofsky · 2022-04-25T16:54:50Z

src/fs.h

+static inline path operator/(path p1, path p2)
 {
-    p1 += std::move(p2);
+    p1 /= std::move(p2);
    return p1;
 }
+static inline path operator/(path p1, std::filesystem::path p2)
+{
+    p1 /= std::move(p2);
+    return p1;
+}
+static inline path operator/(path p1, const char* p2)
+{
+    p1 /= p2;
+    return p1;
+}
+static inline path operator+(path p1, const char* p2)
+{
+    p1 += p2;
+    return p1;
+}
+static inline path operator+(path p1, path::value_type p2)
+{
+    p1 += p2;
+    return p1;
+}


re: #24470 (comment)

Wondering why there's a difference between the allowed right-hand operands for / and +

The cases that are listed are just cases that seemed safe and are in used by existing code. (I think it'd be fine to add cases that are safe and aren't in use by existing code, too, but I didn't go looking for these.)

On the specific cases, combining two path objects with / is safe and useful. Combining two path objects with + is safe probably not useful. Appending a literal string is safe to a path is safe (assuming the literal string is ASCII) both with + and /. Adding a native character to a path is safe and semi-useful (used by some tests at least)

hebasto

ACK f64aa9c

…lowed b…

DrahtBot added GUI P2P Utils/log/libs UTXO Db and Indexes Validation Wallet labels Mar 3, 2022

This was referenced Mar 4, 2022

refactor: Remove gArgs from bdb.h and sqlite.h #23732

Merged

Don't call lsn_reset in periodic flush #18904

Closed

Trinacristella approved these changes Mar 5, 2022

View reviewed changes

ryanofsky mentioned this pull request Mar 7, 2022

Add missing fs::u8path wrap #24493

Closed

DrahtBot added the Needs rebase label Mar 24, 2022

laanwj added Refactoring and removed GUI Wallet UTXO Db and Indexes P2P Validation Utils/log/libs labels Apr 7, 2022

ryanofsky force-pushed the pr/patha branch from 67ca71e to cb9f6b1 Compare April 22, 2022 02:16

DrahtBot removed the Needs rebase label Apr 22, 2022

hebasto reviewed Apr 22, 2022

View reviewed changes

dongcarl reviewed Apr 22, 2022

View reviewed changes

ryanofsky force-pushed the pr/patha branch from cb9f6b1 to f64aa9c Compare April 25, 2022 17:54

ryanofsky commented Apr 25, 2022

View reviewed changes

hebasto approved these changes Apr 25, 2022

View reviewed changes

maflcko merged commit 12455ac into bitcoin:master May 3, 2022

sidhujag pushed a commit to syscoin/syscoin that referenced this pull request May 4, 2022

Merge bitcoin#24470: Disallow more unsafe string->path conversions al…

ad37f1d

…lowed b…

bitcoin locked and limited conversation to collaborators May 3, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Disallow more unsafe string->path conversions allowed by path append operators #24470

Disallow more unsafe string->path conversions allowed by path append operators #24470

Uh oh!

ryanofsky commented Mar 3, 2022

Uh oh!

DrahtBot commented Mar 3, 2022 •

edited

Loading

Uh oh!

hebasto commented Mar 5, 2022

Uh oh!

laanwj commented Apr 7, 2022

Uh oh!

ryanofsky commented Apr 22, 2022

Uh oh!

hebasto commented Apr 22, 2022 •

edited

Loading

Uh oh!

hebasto Apr 22, 2022

Uh oh!

ryanofsky Apr 25, 2022

Uh oh!

dongcarl Apr 22, 2022

Uh oh!

ryanofsky Apr 25, 2022

Uh oh!

ryanofsky left a comment

Uh oh!

ryanofsky Apr 25, 2022

Uh oh!

ryanofsky Apr 25, 2022

Uh oh!

hebasto left a comment

Uh oh!

Uh oh!

Disallow more unsafe string->path conversions allowed by path append operators #24470

Disallow more unsafe string->path conversions allowed by path append operators #24470

Uh oh!

Conversation

ryanofsky commented Mar 3, 2022

Uh oh!

DrahtBot commented Mar 3, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Conflicts

Uh oh!

hebasto commented Mar 5, 2022

Uh oh!

laanwj commented Apr 7, 2022

Uh oh!

ryanofsky commented Apr 22, 2022

Uh oh!

hebasto commented Apr 22, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hebasto Apr 22, 2022

Choose a reason for hiding this comment

Uh oh!

ryanofsky Apr 25, 2022

Choose a reason for hiding this comment

Uh oh!

dongcarl Apr 22, 2022

Choose a reason for hiding this comment

Uh oh!

ryanofsky Apr 25, 2022

Choose a reason for hiding this comment

Uh oh!

ryanofsky left a comment

Choose a reason for hiding this comment

Uh oh!

ryanofsky Apr 25, 2022

Choose a reason for hiding this comment

Uh oh!

ryanofsky Apr 25, 2022

Choose a reason for hiding this comment

Uh oh!

hebasto left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

DrahtBot commented Mar 3, 2022 •

edited

Loading

hebasto commented Apr 22, 2022 •

edited

Loading