[libc] Add shm_open/shm_unlink (llvm#84974)

Author: SchrodingerZhu

libc/config/linux/aarch64/entrypoints.txt

@@ -216,6 +216,8 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.sys.mman.mlockall
     libc.src.sys.mman.munlockall
     libc.src.sys.mman.msync
+    libc.src.sys.mman.shm_open
+    libc.src.sys.mman.shm_unlink
 
     # sys/random.h entrypoints
     libc.src.sys.random.getrandom

libc/config/linux/api.td

@@ -117,7 +117,7 @@ def SchedAPI : PublicAPI<"sched.h"> {
 }
 
 def SysMManAPI : PublicAPI<"sys/mman.h"> {
-  let Types = ["off_t", "size_t"];
+  let Types = ["off_t", "size_t", "mode_t"];
 }
 
 def SignalAPI : PublicAPI<"signal.h"> {

libc/config/linux/riscv/entrypoints.txt

@@ -221,6 +221,8 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.sys.mman.mlockall
     libc.src.sys.mman.munlockall
     libc.src.sys.mman.msync
+    libc.src.sys.mman.shm_open
+    libc.src.sys.mman.shm_unlink
 
     # sys/random.h entrypoints
     libc.src.sys.random.getrandom

libc/config/linux/x86_64/entrypoints.txt

@@ -229,6 +229,8 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.sys.mman.mlockall
     libc.src.sys.mman.munlockall
     libc.src.sys.mman.msync
+    libc.src.sys.mman.shm_open
+    libc.src.sys.mman.shm_unlink
 
     # sys/random.h entrypoints
     libc.src.sys.random.getrandom

libc/docs/dev/undefined_behavior.rst

@@ -70,3 +70,7 @@ Design Decisions
 Resizable Tables for hsearch
 ----------------------------
 The POSIX.1 standard does not delineate the behavior consequent to invoking hsearch or hdestroy without prior initialization of the hash table via hcreate. Furthermore, the standard does not specify the outcomes of successive invocations of hsearch absent intervening hdestroy calls. Libraries such as MUSL and Glibc do not apply checks to these scenarios, potentially leading to memory corruption or leakage. Conversely, FreeBSD's libc and Bionic automatically initialize the hash table to a minimal size if it is found uninitialized, and proceeding to destroy the table only if initialization has occurred. This approach also avoids redundant table allocation if an initialized hash table is already present. Given that the hash table starts with a minimal size, resizing becomes necessary to accommodate additional user insertions. LLVM's libc mirrors the approach of FreeBSD's libc and Bionic, owing to its enhanced robustness and user-friendliness. Notably, such resizing behavior itself aligns with POSIX.1 standards, which explicitly permit implementations to modify the capacity of the hash table.
+
+Path without Leading Slashs in shm_open
+----------------------------------------
+POSIX.1 leaves that when the name of a shared memory object does not begin with a slash, the behavior is implementation defined. In such cases, the shm_open in LLVM libc is implemented to behave as if the name began with a slash.

libc/spec/posix.td

@@ -246,6 +246,7 @@ def POSIX : StandardSpec<"POSIX"> {
       [
         SizeTType,
         OffTType,
+        ModeTType,
       ],
       [], // Enumerations
       [
@@ -310,6 +311,16 @@ def POSIX : StandardSpec<"POSIX"> {
           RetValSpec<IntType>,
           [ArgSpec<VoidPtr>, ArgSpec<SizeTType>, ArgSpec<IntType>]
         >,
+        FunctionSpec<
+          "shm_open",
+          RetValSpec<IntType>,
+          [ArgSpec<ConstCharPtr>, ArgSpec<IntType>, ArgSpec<ModeTType>]
+        >,
+        FunctionSpec<
+          "shm_unlink",
+          RetValSpec<IntType>,
+          [ArgSpec<ConstCharPtr>]
+        >,
       ]
   >;
 

libc/src/__support/CPP/string_view.h

@@ -179,21 +179,38 @@ class string_view {
   LIBC_INLINE char back() const { return Data[Len - 1]; }
 
   // Finds the first occurence of c in this view, starting at position From.
-  LIBC_INLINE size_t find_first_of(const char c, size_t From = 0) const {
+  LIBC_INLINE constexpr size_t find_first_of(const char c,
+                                             size_t From = 0) const {
     for (size_t Pos = From; Pos < size(); ++Pos)
       if ((*this)[Pos] == c)
         return Pos;
     return npos;
   }
 
   // Finds the last occurence of c in this view, ending at position End.
-  LIBC_INLINE size_t find_last_of(const char c, size_t End = npos) const {
+  LIBC_INLINE constexpr size_t find_last_of(const char c,
+                                            size_t End = npos) const {
     End = End >= size() ? size() : End + 1;
     for (; End > 0; --End)
       if ((*this)[End - 1] == c)
         return End - 1;
     return npos;
   }
+
+  // Finds the first character not equal to c in this view, starting at position
+  // From.
+  LIBC_INLINE constexpr size_t find_first_not_of(const char c,
+                                                 size_t From = 0) const {
+    for (size_t Pos = From; Pos < size(); ++Pos)
+      if ((*this)[Pos] != c)
+        return Pos;
+    return npos;
+  }
+
+  // Check if this view contains the given character.
+  LIBC_INLINE constexpr bool contains(char c) const {
+    return find_first_of(c) != npos;
+  }
 };
 
 } // namespace cpp

libc/src/sys/mman/CMakeLists.txt

@@ -85,3 +85,17 @@ add_entrypoint_object(
   DEPENDS
     .${LIBC_TARGET_OS}.msync
 )
+
+add_entrypoint_object(
+  shm_open
+  ALIAS
+  DEPENDS
+    .${LIBC_TARGET_OS}.shm_open
+)
+
+add_entrypoint_object(
+  shm_unlink
+  ALIAS
+  DEPENDS
+    .${LIBC_TARGET_OS}.shm_unlink
+)

libc/src/sys/mman/linux/CMakeLists.txt

@@ -152,3 +152,40 @@ add_entrypoint_object(
     libc.src.__support.OSUtil.osutil
     libc.src.errno.errno
 )
+
+add_header_library(
+  shm_common
+  HDRS
+    shm_common.h
+  DEPENDS
+    libc.src.__support.CPP.array
+    libc.src.__support.CPP.string_view
+    libc.src.__support.CPP.optional
+    libc.src.__support.common
+    libc.src.errno.errno
+    libc.src.string.memory_utils.inline_memcpy
+)
+
+add_entrypoint_object(
+  shm_open
+  SRCS
+    shm_open.cpp
+  HDRS
+    ../shm_open.h
+  DEPENDS
+    libc.src.fcntl.open
+    libc.include.llvm-libc-macros.fcntl_macros
+    libc.include.llvm-libc-types.mode_t
+    .shm_common
+)
+
+add_entrypoint_object(
+  shm_unlink
+  SRCS
+    shm_unlink.cpp
+  HDRS
+    ../shm_unlink.h
+  DEPENDS
+    libc.src.unistd.unlink
+    .shm_common
+)

libc/src/sys/mman/linux/shm_common.h

@@ -0,0 +1,53 @@
+//===---------- Shared implementations for shm_open/shm_unlink ------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/__support/CPP/array.h"
+#include "src/__support/CPP/optional.h"
+#include "src/__support/CPP/string_view.h"
+#include "src/errno/libc_errno.h"
+#include "src/string/memory_utils/inline_memcpy.h"
+
+// TODO: Get PATH_MAX via https://github.com/llvm/llvm-project/issues/85121
+#include <linux/limits.h>
+
+namespace LIBC_NAMESPACE {
+
+namespace shm_common {
+
+LIBC_INLINE_VAR constexpr cpp::string_view SHM_PREFIX = "/dev/shm/";
+using SHMPath = cpp::array<char, NAME_MAX + SHM_PREFIX.size() + 1>;
+
+LIBC_INLINE cpp::optional<SHMPath> translate_name(cpp::string_view name) {
+  // trim leading slashes
+  size_t offset = name.find_first_not_of('/');
+  if (offset == cpp::string_view::npos) {
+    libc_errno = EINVAL;
+    return cpp::nullopt;
+  }
+  name = name.substr(offset);
+
+  // check the name
+  if (name.size() > NAME_MAX) {
+    libc_errno = ENAMETOOLONG;
+    return cpp::nullopt;
+  }
+  if (name == "." || name == ".." || name.contains('/')) {
+    libc_errno = EINVAL;
+    return cpp::nullopt;
+  }
+
+  // prepend the prefix
+  SHMPath buffer;
+  inline_memcpy(buffer.data(), SHM_PREFIX.data(), SHM_PREFIX.size());
+  inline_memcpy(buffer.data() + SHM_PREFIX.size(), name.data(), name.size());
+  buffer[SHM_PREFIX.size() + name.size()] = '\0';
+  return buffer;
+}
+} // namespace shm_common
+
+} // namespace LIBC_NAMESPACE