Turn on `CONFIG_WERROR` for x86_64 #245

nathanchance · 2021-11-19T00:16:14Z

To turn this on, we need allmodconfig and allyesconfig to build without any warnings on linux-5.15.y, mainline, and -next with LLVM 11 through 14 (quite the matrix).

Current status as of November 18, 2021

linux-5.15.y @ v5.15.3:

LLVM 11: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3304 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
           ^
1 error generated.

LLVM 12: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3288 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
          ^
1 error generated.

LLVM 13: PASS
LLVM 14: FAIL

drivers/power/reset/ltc2952-poweroff.c:164:28: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->wde_interval = 300L * 1E6L;
                                ^
drivers/power/reset/ltc2952-poweroff.c:164:21: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->wde_interval = 300L * 1E6L;
                         ^
drivers/power/reset/ltc2952-poweroff.c:165:41: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->trigger_delay = ktime_set(2, 500L*1E6L);
                                             ^
3 errors generated.

drivers/usb/dwc2/hcd_queue.c:1744:25: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
                      delay = ktime_set(0, DWC2_RETRY_WAIT_DELAY);
                                           ^
drivers/usb/dwc2/hcd_queue.c:62:34: note: expanded from macro 'DWC2_RETRY_WAIT_DELAY'
#define DWC2_RETRY_WAIT_DELAY (1 * 1E6L)
                               ^
1 error generated.

drivers/gpu/drm/i915/intel_pm.c:3066:12: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
        changed = ilk_increase_wm_latency(dev_priv, dev_priv->wm.pri_latency, 12) |
                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/i915/intel_pm.c:3066:12: note: cast one or both operands to int to silence this warning
drivers/gpu/drm/i915/intel_pm.c:3066:12: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
        changed = ilk_increase_wm_latency(dev_priv, dev_priv->wm.pri_latency, 12) |
                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                                                                ||
drivers/gpu/drm/i915/intel_pm.c:3066:12: note: cast one or both operands to int to silence this warning
2 errors generated.

In file included from drivers/staging/wlan-ng/prism2usb.c:2:
drivers/staging/wlan-ng/hfa384x_usb.c:3787:7: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
          ((test_and_clear_bit(THROTTLE_RX, &hw->usb_flags) &&
          ~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/staging/wlan-ng/hfa384x_usb.c:3787:7: note: cast one or both operands to int to silence this warning
1 error generated.

Mainline @ 42eb8fdac2fc5d62392dcfcf0253753e821a97b0:

LLVM 11: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3304 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
           ^
1 error generated.

LLVM 12: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3288 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
          ^
1 error generated.

LLVM 13: PASS
LLVM 14: FAIL

drivers/gpu/drm/i915/intel_pm.c:3065:12: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
        changed = ilk_increase_wm_latency(dev_priv, dev_priv->wm.pri_latency, 12) |
                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/i915/intel_pm.c:3065:12: note: cast one or both operands to int to silence this warning
drivers/gpu/drm/i915/intel_pm.c:3065:12: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
        changed = ilk_increase_wm_latency(dev_priv, dev_priv->wm.pri_latency, 12) |
                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                                                                                  ||
drivers/gpu/drm/i915/intel_pm.c:3065:12: note: cast one or both operands to int to silence this warning
2 errors generated.

drivers/power/reset/ltc2952-poweroff.c:162:28: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
        data->wde_interval = 300L * 1E6L;
                                  ^
drivers/power/reset/ltc2952-poweroff.c:162:21: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
        data->wde_interval = 300L * 1E6L;
                         ^
drivers/power/reset/ltc2952-poweroff.c:163:41: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
        data->trigger_delay = ktime_set(2, 500L*1E6L);
                                               ^
3 errors generated.

drivers/usb/dwc2/hcd_queue.c:1744:25: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
                        delay = ktime_set(0, DWC2_RETRY_WAIT_DELAY);
                                             ^
drivers/usb/dwc2/hcd_queue.c:62:34: note: expanded from macro 'DWC2_RETRY_WAIT_DELAY'
#define DWC2_RETRY_WAIT_DELAY (1 * 1E6L)
                                 ^
1 error generated.

-next @ next-20211118:

LLVM 11: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3304 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
           ^
1 error generated.

LLVM 12: FAIL

drivers/staging/greybus/audio_topology.c:977:12: error: stack frame size of 3288 bytes in function 'gbaudio_tplg_create_widget' [-Werror,-Wframe-larger-than=]
static int gbaudio_tplg_create_widget(struct gbaudio_module_info *module,
          ^
1 error generated.

LLVM 13: PASS
LLVM 14: FAIL

drivers/power/reset/ltc2952-poweroff.c:162:28: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->wde_interval = 300L * 1E6L;
                                ^
drivers/power/reset/ltc2952-poweroff.c:162:21: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->wde_interval = 300L * 1E6L;
                         ^
drivers/power/reset/ltc2952-poweroff.c:163:41: error: expression requires  'long double' type support, but target 'x86_64-unknown-linux-gnu' does not support it
      data->trigger_delay = ktime_set(2, 500L*1E6L);
                                             ^
3 errors generated.

Analysis:

drivers/staging/greybus/audio_topology.c: @arndb sent a patch back in January with some feedback but I never saw a respin.
drivers/gpu/drm/i915/intel_pm.c: I sent a patch, which has been accepted but it missed the 5.16 merge window, meaning it will not be fixed until 5.17 :(
drivers/power/reset/ltc2952-poweroff.c: I sent a patch, which has been accepted in a fixes branch, meaning it should be merged into mainline soon.
drivers/usb/dwc2/hcd_queue.c: I sent a patch, which has been accepted in a fixes branch, meaning it should be merged into mainline soon.
drivers/staging/wlan-ng/hfa384x_usb.c: The fix is available in 5.16-rc1, it just needs to be backported to stable.

The text was updated successfully, but these errors were encountered:

nathanchance · 2021-12-10T00:20:06Z

Mainline status as of 12/9/2021

I am going to start posting updates here against mainline, as -next is a moving target and stable will be fixable based on mainline's status. Android has disabled CONFIG_WERROR as well so this should be high priority for us to fix.

Outstanding warnings with allmodconfig:

-Wbitwise-instead-of-logical in drivers/gpu/drm/i915/intel_pm.c linux#1473 (clang-14 and newer): The patch is accepted and in -next. It will be merged in 5.17, after which it can be backported to 5.15 and 5.16.
-Wframe-larger-than= in drivers/staging/greybus/audio_topology.c linux#1535 (clang-11 and clang-12): I sent a v2 for review. Hopefully, this can be merged for 5.17.

nathanchance · 2022-01-13T22:52:06Z

As of fb3b0673b7d5 ("Merge tag 'mailbox-v5.17' of git://git.linaro.org/landing-teams/working/fujitsu/integration"), ARCH=x86_64 allmodconfig is warning free.

I requested backports to 5.15 and 5.16, which will allow us to enable CONFIG_WERROR=y for these trees.

However, linux-next is not warning free, due to 743b237c3a7b ("scsi: storvsc: Add Isolation VM support for storvsc driver"), which adds a false positive instance of -Wshift-count-overflow, so this is now blocked on fixing that in the compiler and adding a workaround for older clang versions.

nathanchance · 2022-02-08T21:56:06Z

Stable trees

5.15.21 and 5.16.7 remain warning free.

Mainline

5.17-rc3 has one warning:

drivers/hv/vmbus_drv.c:2082:29: error: shift count >= width of type [-Werror,-Wshift-count-overflow]
static u64 vmbus_dma_mask = DMA_BIT_MASK(64);
                            ^~~~~~~~~~~~~~~~
./include/linux/dma-mapping.h:76:54: note: expanded from macro 'DMA_BIT_MASK'
#define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
                                                     ^ ~~~
1 error generated.

Fixed with https://git.kernel.org/hyperv/c/6bf625a4140f24b490766043b307f8252519578b, which will hopefully be in 5.17-rc4. This patch works around the compiler bug that I mention in my comment above, which allows us to remove the "blocked (llvm)" label.

-next

On next-20220208:

fs/io_uring.c:9396:9: error: variable 'ret' is uninitialized when used here [-Werror,-Wuninitialized]
        return ret;
               ^~~
fs/io_uring.c:9373:13: note: initialize the variable 'ret' to silence this warning
        int fd, ret;
                   ^
                    = 0
1 error generated.

Fixed by https://git.kernel.org/axboe/linux-block/c/4c65723081332607ca331072b0f8a90189e2e447, which should be in next-20220209.

There are also a few warnings from the new FORTIFY_SOURCE compile time checks with clang-14 and clang-15:

In file included from drivers/scsi/libfc/fc_elsct.c:17:
In file included from ./include/scsi/libfc.h:11:
In file included from ./include/linux/timer.h:6:
In file included from ./include/linux/ktime.h:24:
In file included from ./include/linux/time.h:60:
In file included from ./include/linux/time32.h:13:
In file included from ./include/linux/timex.h:65:
In file included from ./arch/arm64/include/asm/timex.h:8:
In file included from ./arch/arm64/include/asm/arch_timer.h:18:
In file included from ./include/linux/smp.h:13:
In file included from ./include/linux/cpumask.h:12:
In file included from ./include/linux/bitmap.h:11:
In file included from ./include/linux/string.h:253:
./include/linux/fortify-string.h:328:4: error: call to __write_overflow_field declared with 'warning' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror,-Wattribute-warning]
                        __write_overflow_field(p_size_field, size);
                        ^
./include/linux/fortify-string.h:328:4: error: call to __write_overflow_field declared with 'warning' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror,-Wattribute-warning]
2 errors generated.

In file included from drivers/misc/habanalabs/common/firmware_if.c:8:
In file included from drivers/misc/habanalabs/common/habanalabs.h:11:
In file included from drivers/misc/habanalabs/common/../include/common/cpucp_if.h:12:
In file included from ./include/linux/if_ether.h:19:
In file included from ./include/linux/skbuff.h:15:
In file included from ./include/linux/time.h:60:
In file included from ./include/linux/time32.h:13:
In file included from ./include/linux/timex.h:65:
In file included from ./arch/arm64/include/asm/timex.h:8:
In file included from ./arch/arm64/include/asm/arch_timer.h:18:
In file included from ./include/linux/smp.h:13:
In file included from ./include/linux/cpumask.h:12:
In file included from ./include/linux/bitmap.h:11:
In file included from ./include/linux/string.h:253:
./include/linux/fortify-string.h:242:4: error: call to __write_overflow_field declared with 'warning' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror,-Wattribute-warning]
                        __write_overflow_field(p_size_field, size);
                        ^
1 error generated.

In file included from drivers/net/ethernet/huawei/hinic/hinic_devlink.c:15:
In file included from ./include/linux/netlink.h:7:
In file included from ./include/linux/skbuff.h:15:
In file included from ./include/linux/time.h:60:
In file included from ./include/linux/time32.h:13:
In file included from ./include/linux/timex.h:65:
In file included from ./arch/arm64/include/asm/timex.h:8:
In file included from ./arch/arm64/include/asm/arch_timer.h:18:
In file included from ./include/linux/smp.h:13:
In file included from ./include/linux/cpumask.h:12:
In file included from ./include/linux/bitmap.h:11:
In file included from ./include/linux/string.h:253:
./include/linux/fortify-string.h:328:4: error: call to __write_overflow_field declared with 'warning' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror,-Wattribute-warning]
                        __write_overflow_field(p_size_field, size);
                        ^
1 error generated.

x86_64 all{mod,yes}config are now -Werror clean on 5.15, stable (5.18), mainline (5.19-rc5), and next-20220705. Turn this on so we can catch new warnings easier. Closes: ClangBuiltLinux#245 Signed-off-by: Nathan Chancellor <[email protected]>

nathanchance added enhancement New feature or request WIP Work in progress labels Nov 19, 2021

nathanchance self-assigned this Nov 19, 2021

nathanchance added the blocked (linux) Blocked on an issue with Linux upstream label Nov 19, 2021

nathanchance mentioned this issue Dec 10, 2021

Turn on CONFIG_WERROR for arm64 #246

Closed

nathanchance added the blocked (llvm) Blocked on an issue with LLVM upstream label Jan 13, 2022

nathanchance removed the blocked (llvm) Blocked on an issue with LLVM upstream label Feb 8, 2022

nathanchance mentioned this issue Jul 5, 2022

Stop disabling CONFIG_WERROR on x86_64 #385

Merged

nathanchance closed this as completed in #385 Jul 6, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Turn on `CONFIG_WERROR` for x86_64 #245

Turn on `CONFIG_WERROR` for x86_64 #245

nathanchance commented Nov 19, 2021

nathanchance commented Dec 10, 2021

Uh oh!

nathanchance commented Jan 13, 2022

Uh oh!

nathanchance commented Feb 8, 2022

Uh oh!

Turn on CONFIG_WERROR for x86_64 #245

Turn on CONFIG_WERROR for x86_64 #245

Comments

nathanchance commented Nov 19, 2021

Current status as of November 18, 2021

Analysis:

nathanchance commented Dec 10, 2021

Mainline status as of 12/9/2021

Uh oh!

nathanchance commented Jan 13, 2022

Uh oh!

nathanchance commented Feb 8, 2022

Stable trees

Mainline

-next

Uh oh!

Turn on `CONFIG_WERROR` for x86_64 #245

Turn on `CONFIG_WERROR` for x86_64 #245