In sgx_init(), if misc_register() fails or misc_register() succeeds but neither sgx_drv_init() nor sgx_vepc_init() succeeds, then ksgxd will be prematurely stopped. This may leave some unsanitized pages, which does not matter, because SGX will be disabled for the whole power cycle. This triggers WARN_ON() because sgx_dirty_page_list ends up being non-empty, and dumps the call stack: [ 0.268103] sgx: EPC section 0x40200000-0x45f7ffff [ 0.268591] ------------[ cut here ]------------ [ 0.268592] WARNING: CPU: 6 PID: 83 at arch/x86/kernel/cpu/sgx/main.c:401 ksgxd+0x1b7/0x1d0 [ 0.268598] Modules linked in: [ 0.268600] CPU: 6 PID: 83 Comm: ksgxd Not tainted 6.0.0-rc2 #382 [ 0.268603] Hardware name: Dell Inc. XPS 13 9370/0RMYH9, BIOS 1.21.0 07/06/2022 [ 0.268604] RIP: 0010:ksgxd+0x1b7/0x1d0 [ 0.268607] Code: ff e9 f2 fe ff ff 48 89 df e8 75 07 0e 00 84 c0 0f 84 c3 fe ff ff 31 ff e8 e6 07 0e 00 84 c0 0f 85 94 fe ff ff e9 af fe ff ff <0f> 0b e9 7f fe ff ff e8 dd 9c 95 00 66 66 2e 0f 1f 84 00 00 00 00 [ 0.268608] RSP: 0000:ffffb6c7404f3ed8 EFLAGS: 00010287 [ 0.268610] RAX: ffffb6c740431a10 RBX: ffff8dcd8117b400 RCX: 0000000000000000 [ 0.268612] RDX: 0000000080000000 RSI: ffffb6c7404319d0 RDI: 00000000ffffffff [ 0.268613] RBP: ffff8dcd820a4d80 R08: ffff8dcd820a4180 R09: ffff8dcd820a4180 [ 0.268614] R10: 0000000000000000 R11: 0000000000000006 R12: ffffb6c74006bce0 [ 0.268615] R13: ffff8dcd80e63880 R14: ffffffffa8a60f10 R15: 0000000000000000 [ 0.268616] FS: 0000000000000000(0000) GS:ffff8dcf25580000(0000) knlGS:0000000000000000 [ 0.268617] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 0.268619] CR2: 0000000000000000 CR3: 0000000213410001 CR4: 00000000003706e0 [ 0.268620] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 0.268621] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 0.268622] Call Trace: [ 0.268624] <TASK> [ 0.268627] ? _raw_spin_lock_irqsave+0x24/0x60 [ 0.268632] ? _raw_spin_unlock_irqrestore+0x23/0x40 [ 0.268634] ? __kthread_parkme+0x36/0x90 [ 0.268637] kthread+0xe5/0x110 [ 0.268639] ? kthread_complete_and_exit+0x20/0x20 [ 0.268642] ret_from_fork+0x1f/0x30 [ 0.268647] </TASK> [ 0.268648] ---[ end trace 0000000000000000 ]--- Ultimately this can crash the kernel, if the following is set: /proc/sys/kernel/panic_on_warn In premature stop, print nothing, as the number is by practical means a random number. Otherwise, it is an indicator of a bug in the driver, and therefore print the number of unsanitized pages with pr_err(). Link: https://lore.kernel.org/linux-sgx/20220825051827.246698-1-jarkko@xxxxxxxxxx/T/#u Fixes: 51ab30eb2ad4 ("x86/sgx: Replace section->init_laundry_list with sgx_dirty_page_list") Cc: stable@xxxxxxxxxxxxxxx # v5.13+ Reported-by: Paul Menzel <pmenzel@xxxxxxxxxxxxx> Signed-off-by: Jarkko Sakkinen <jarkko@xxxxxxxxxx> --- v6: - Address Reinette's feedback: https://lore.kernel.org/linux-sgx/Yw6%2FiTzSdSw%2FY%2FVO@xxxxxxxxxx/ v5: - Add the klog dump and sysctl option to the commit message. v4: - Explain expectations for dirty_page_list in the function header, instead of an inline comment. - Improve commit message to explain the conditions better. - Return the number of pages left dirty to ksgxd() and print warning after the 2nd call, if there are any. v3: - Remove WARN_ON(). - Tuned comments and the commit message a bit. v2: - Replaced WARN_ON() with optional pr_info() inside __sgx_sanitize_pages(). - Rewrote the commit message. - Added the fixes tag. --- arch/x86/kernel/cpu/sgx/main.c | 42 ++++++++++++++++++++++++++++------ 1 file changed, 35 insertions(+), 7 deletions(-) diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c index 515e2a5f25bb..bcd6b64961bd 100644 --- a/arch/x86/kernel/cpu/sgx/main.c +++ b/arch/x86/kernel/cpu/sgx/main.c @@ -49,17 +49,20 @@ static LIST_HEAD(sgx_dirty_page_list); * Reset post-kexec EPC pages to the uninitialized state. The pages are removed * from the input list, and made available for the page allocator. SECS pages * prepending their children in the input list are left intact. + * + * Contents of the @dirty_page_list must be thread-local, i.e. + * not shared by multiple threads. */ -static void __sgx_sanitize_pages(struct list_head *dirty_page_list) +static long __sgx_sanitize_pages(struct list_head *dirty_page_list) { struct sgx_epc_page *page; + long left_dirty = 0; LIST_HEAD(dirty); int ret; - /* dirty_page_list is thread-local, no need for a lock: */ while (!list_empty(dirty_page_list)) { if (kthread_should_stop()) - return; + return -ECANCELED; page = list_first_entry(dirty_page_list, struct sgx_epc_page, list); @@ -92,12 +95,14 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list) } else { /* The page is not yet clean - move to the dirty list. */ list_move_tail(&page->list, &dirty); + left_dirty++; } cond_resched(); } list_splice(&dirty, dirty_page_list); + return left_dirty; } static bool sgx_reclaimer_age(struct sgx_epc_page *epc_page) @@ -388,17 +393,40 @@ void sgx_reclaim_direct(void) static int ksgxd(void *p) { + long ret; + set_freezable(); /* * Sanitize pages in order to recover from kexec(). The 2nd pass is * required for SECS pages, whose child pages blocked EREMOVE. */ - __sgx_sanitize_pages(&sgx_dirty_page_list); - __sgx_sanitize_pages(&sgx_dirty_page_list); + ret = __sgx_sanitize_pages(&sgx_dirty_page_list); + if (ret == -ECANCELED) + /* kthread stopped */ + return 0; - /* sanity check: */ - WARN_ON(!list_empty(&sgx_dirty_page_list)); + ret = __sgx_sanitize_pages(&sgx_dirty_page_list); + switch (ret) { + case 0: + /* success, no unsanitized pages */ + break; + + case -ECANCELED: + /* kthread stopped */ + return 0; + + default: + /* + * Never expected to happen in a working driver. If it happens + * the bug is expected to be in the sanitization process, but + * successfully sanitized pages are still valid and driver can + * be used and most importantly debugged without issues. To put + * short, the global state of kernel is not corrupted so no + * reason to do any more complicated rollback. + */ + pr_err("%ld unsanitized pages\n", ret); + } while (!kthread_should_stop()) { if (try_to_freeze()) -- 2.37.2
