public inbox for linux-mm@kvack.org
 help / color / mirror / Atom feed
* [RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata
@ 2026-03-27 13:22 Dmitry Antipov
  2026-03-31 13:37 ` kernel test robot
  0 siblings, 1 reply; 2+ messages in thread
From: Dmitry Antipov @ 2026-03-27 13:22 UTC (permalink / raw)
  To: Luis Chamberlain, Petr Pavlu, Daniel Gomez, Sami Tolvanen
  Cc: linux-modules, linux-mm, Dmitry Antipov

Since kernel modules has their own .rodata sections, functions like
'kstrdup_const()' called from the module context are not required to
copy string constants from these sections. Likewise, 'kfree_const()'
in such a context becomes a no-op also because the whole module's
.rodata is freed at module unloading. OTOH this proof-of-concept
implementation introduces substantial overhead due to calls to
'__module_address()' from 'is_module_rodata()', which is much slower
than 'is_kernel_rodata()'. Anyway, comments are highly appreciated.

Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
---
 include/linux/module.h | 6 ++++++
 kernel/module/main.c   | 9 +++++++++
 mm/util.c              | 7 +++++--
 3 files changed, 20 insertions(+), 2 deletions(-)

diff --git a/include/linux/module.h b/include/linux/module.h
index 14f391b186c6..7edaf2b730ce 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -623,6 +623,7 @@ bool is_module_address(unsigned long addr);
 bool __is_module_percpu_address(unsigned long addr, unsigned long *can_addr);
 bool is_module_percpu_address(unsigned long addr);
 bool is_module_text_address(unsigned long addr);
+bool is_module_rodata(unsigned long addr);
 
 static inline bool within_module_mem_type(unsigned long addr,
 					  const struct module *mod,
@@ -807,6 +808,11 @@ static inline bool is_module_text_address(unsigned long addr)
 	return false;
 }
 
+static inline bool is_module_rodata(unsigned long addr)
+{
+	return false;
+}
+
 static inline bool within_module_core(unsigned long addr,
 				      const struct module *mod)
 {
diff --git a/kernel/module/main.c b/kernel/module/main.c
index c3ce106c70af..5003cacd0786 100644
--- a/kernel/module/main.c
+++ b/kernel/module/main.c
@@ -3858,6 +3858,15 @@ bool is_module_text_address(unsigned long addr)
 	return __module_text_address(addr) != NULL;
 }
 
+bool is_module_rodata(unsigned long addr)
+{
+	struct module *mod;
+
+	guard(rcu)();
+	mod = __module_address(addr);
+	return mod && within_module_mem_type(addr, mod, MOD_RODATA);
+}
+
 void module_for_each_mod(int(*func)(struct module *mod, void *data), void *data)
 {
 	struct module *mod;
diff --git a/mm/util.c b/mm/util.c
index b05ab6f97e11..8dd1f1e95554 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -26,6 +26,7 @@
 #include <linux/compat.h>
 #include <linux/fsnotify.h>
 #include <linux/page_idle.h>
+#include <linux/module.h>
 
 #include <linux/uaccess.h>
 
@@ -42,7 +43,8 @@
  */
 void kfree_const(const void *x)
 {
-	if (!is_kernel_rodata((unsigned long)x))
+	if (!is_kernel_rodata((unsigned long)x) &&
+	    !is_module_rodata((unsigned long)x))
 		kfree(x);
 }
 EXPORT_SYMBOL(kfree_const);
@@ -98,7 +100,8 @@ EXPORT_SYMBOL(kstrdup);
  */
 const char *kstrdup_const(const char *s, gfp_t gfp)
 {
-	if (is_kernel_rodata((unsigned long)s))
+	if (is_kernel_rodata((unsigned long)s) ||
+	    is_module_rodata((unsigned long)s))
 		return s;
 
 	return kstrdup(s, gfp);
-- 
2.53.0



^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata
  2026-03-27 13:22 [RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata Dmitry Antipov
@ 2026-03-31 13:37 ` kernel test robot
  0 siblings, 0 replies; 2+ messages in thread
From: kernel test robot @ 2026-03-31 13:37 UTC (permalink / raw)
  To: Dmitry Antipov
  Cc: oe-lkp, lkp, linux-modules, linux-kernel, linux-mm,
	Luis Chamberlain, Petr Pavlu, Daniel Gomez, Sami Tolvanen,
	Dmitry Antipov, oliver.sang



Hello,

kernel test robot noticed "kernel_BUG_at_arch/x86/mm/physaddr.c" on:

commit: 0803ad97505a632fc3d0cddd6659617deb0ff34b ("[RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata")
url: https://github.com/intel-lab-lkp/linux/commits/Dmitry-Antipov/modules-extend-kstrdup-kfree-_const-to-handle-per-module-rodata/20260329-123400
base: https://git.kernel.org/cgit/linux/kernel/git/akpm/mm.git mm-everything
patch link: https://lore.kernel.org/all/20260327132247.861984-1-dmantipov@yandex.ru/
patch subject: [RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata

in testcase: boot

config: i386-randconfig-002-20260331
compiler: gcc-14
test machine: qemu-system-i386 -enable-kvm -cpu SandyBridge -smp 2 -m 4G

(please refer to attached dmesg/kmsg for entire log/backtrace)



If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <oliver.sang@intel.com>
| Closes: https://lore.kernel.org/oe-lkp/202603312130.f9ed8d89-lkp@intel.com



[   17.209008][    C0] ------------[ cut here ]------------
[   17.209572][    C0] kernel BUG at arch/x86/mm/physaddr.c:66!
[   17.210058][    C0] Oops: invalid opcode: 0000 [#1]
[   17.210470][    C0] CPU: 0 UID: 0 PID: 13 Comm: ksoftirqd/0 Not tainted 7.0.0-rc5-00667-g0803ad97505a #1 PREEMPT(full)  4fef6f10f7892f9689059524fd4d050b8163a9be
[   17.211614][    C0] EIP: __phys_addr (arch/x86/mm/physaddr.c:66 (discriminator 3))
[   17.211995][    C0] Code: e5 56 53 85 c0 78 09 0f 0b b8 ec a1 e6 81 eb 21 80 3d f8 e9 4d 82 00 89 c3 8d b0 00 00 00 80 74 15 e8 cf c5 19 00 84 c0 74 0c <0f> 0b b8 e0 a1 e6 81 e8 77 3e fb ff a1 30 14 d3 82 85 c0 74 2a 89
All code
========
   0:	e5 56                	in     $0x56,%eax
   2:	53                   	push   %rbx
   3:	85 c0                	test   %eax,%eax
   5:	78 09                	js     0x10
   7:	0f 0b                	ud2
   9:	b8 ec a1 e6 81       	mov    $0x81e6a1ec,%eax
   e:	eb 21                	jmp    0x31
  10:	80 3d f8 e9 4d 82 00 	cmpb   $0x0,-0x7db21608(%rip)        # 0xffffffff824dea0f
  17:	89 c3                	mov    %eax,%ebx
  19:	8d b0 00 00 00 80    	lea    -0x80000000(%rax),%esi
  1f:	74 15                	je     0x36
  21:	e8 cf c5 19 00       	call   0x19c5f5
  26:	84 c0                	test   %al,%al
  28:	74 0c                	je     0x36
  2a:*	0f 0b                	ud2		<-- trapping instruction
  2c:	b8 e0 a1 e6 81       	mov    $0x81e6a1e0,%eax
  31:	e8 77 3e fb ff       	call   0xfffffffffffb3ead
  36:	a1 30 14 d3 82 85 c0 	movabs 0x2a74c08582d31430,%eax
  3d:	74 2a 
  3f:	89                   	.byte 0x89

Code starting with the faulting instruction
===========================================
   0:	0f 0b                	ud2
   2:	b8 e0 a1 e6 81       	mov    $0x81e6a1e0,%eax
   7:	e8 77 3e fb ff       	call   0xfffffffffffb3e83
   c:	a1 30 14 d3 82 85 c0 	movabs 0x2a74c08582d31430,%eax
  13:	74 2a 
  15:	89                   	.byte 0x89
[   17.214426][    C0] EAX: f7e9f401 EBX: f7e9f4ac ECX: fffff000 EDX: ff5fe000
[   17.215476][    C0] ESI: 77e9f4ac EDI: 00000000 EBP: 80b0be68 ESP: 80b0be60
[   17.216539][    C0] DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068 EFLAGS: 00010202
[   17.217741][    C0] CR0: 80050033 CR2: 7fbf9dd8 CR3: 032cd000 CR4: 000406b0
[   17.218849][    C0] Call Trace:
[   17.219371][    C0]  kfree (mm/slub.c:6474 (discriminator 1))
[   17.220011][    C0]  ? rcu_read_unlock (include/linux/rcupdate.h:884)
[   17.220786][    C0]  kfree_const (mm/util.c:49)
[   17.221549][    C0]  kernfs_free_rcu (fs/kernfs/dir.c:550)
[   17.222201][    C0]  ? lock_is_held (arch/x86/events/intel/core.c:4034)
[   17.222920][    C0]  rcu_do_batch+0x329/0x6a8
[   17.223812][    C0]  rcu_core (kernel/rcu/tree.c:2871)
[   17.224493][    C0]  rcu_core_si (kernel/rcu/tree.c:2887)
[   17.225255][    C0]  handle_softirqs (arch/x86/include/asm/atomic.h:23 include/linux/atomic/atomic-arch-fallback.h:457 include/linux/jump_label.h:262 include/trace/events/irq.h:142 kernel/softirq.c:623)
[   17.226002][    C0]  run_ksoftirqd (kernel/softirq.c:479 kernel/softirq.c:1064 kernel/softirq.c:1055)
[   17.226737][    C0]  smpboot_thread_fn (kernel/smpboot.c:160)
[   17.227510][    C0]  ? sort_range (kernel/smpboot.c:103)
[   17.228198][    C0]  kthread (kernel/kthread.c:438)
[   17.228839][    C0]  ? sort_range (kernel/smpboot.c:103)
[   17.229601][    C0]  ? kthread_affine_node (kernel/kthread.c:381)
[   17.230400][    C0]  ret_from_fork (arch/x86/kernel/process.c:164)
[   17.231112][    C0]  ? kthread_affine_node (kernel/kthread.c:381)
[   17.231971][    C0]  ret_from_fork_asm (arch/x86/entry/entry_32.S:737)
[   17.232755][    C0]  entry_INT80_32 (arch/x86/entry/entry_32.S:942)
[   17.233621][    C0] Modules linked in: usbtest(+) serio_raw evdev parport_pc parport qemu_fw_cfg button
[   17.235190][    C0] ---[ end trace 0000000000000000 ]---
[   17.236068][    C0] EIP: __phys_addr (arch/x86/mm/physaddr.c:66 (discriminator 3))
[   17.236831][    C0] Code: e5 56 53 85 c0 78 09 0f 0b b8 ec a1 e6 81 eb 21 80 3d f8 e9 4d 82 00 89 c3 8d b0 00 00 00 80 74 15 e8 cf c5 19 00 84 c0 74 0c <0f> 0b b8 e0 a1 e6 81 e8 77 3e fb ff a1 30 14 d3 82 85 c0 74 2a 89
All code
========
   0:	e5 56                	in     $0x56,%eax
   2:	53                   	push   %rbx
   3:	85 c0                	test   %eax,%eax
   5:	78 09                	js     0x10
   7:	0f 0b                	ud2
   9:	b8 ec a1 e6 81       	mov    $0x81e6a1ec,%eax
   e:	eb 21                	jmp    0x31
  10:	80 3d f8 e9 4d 82 00 	cmpb   $0x0,-0x7db21608(%rip)        # 0xffffffff824dea0f
  17:	89 c3                	mov    %eax,%ebx
  19:	8d b0 00 00 00 80    	lea    -0x80000000(%rax),%esi
  1f:	74 15                	je     0x36
  21:	e8 cf c5 19 00       	call   0x19c5f5
  26:	84 c0                	test   %al,%al
  28:	74 0c                	je     0x36
  2a:*	0f 0b                	ud2		<-- trapping instruction
  2c:	b8 e0 a1 e6 81       	mov    $0x81e6a1e0,%eax
  31:	e8 77 3e fb ff       	call   0xfffffffffffb3ead
  36:	a1 30 14 d3 82 85 c0 	movabs 0x2a74c08582d31430,%eax
  3d:	74 2a 
  3f:	89                   	.byte 0x89

Code starting with the faulting instruction
===========================================
   0:	0f 0b                	ud2
   2:	b8 e0 a1 e6 81       	mov    $0x81e6a1e0,%eax
   7:	e8 77 3e fb ff       	call   0xfffffffffffb3e83
   c:	a1 30 14 d3 82 85 c0 	movabs 0x2a74c08582d31430,%eax
  13:	74 2a 
  15:	89                   	.byte 0x89


The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20260331/202603312130.f9ed8d89-lkp@intel.com



-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-03-31 13:37 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-03-27 13:22 [RFC PATCH] modules: extend {kstrdup,kfree}_const() to handle per-module .rodata Dmitry Antipov
2026-03-31 13:37 ` kernel test robot

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox