lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20230815165827.GG907732@google.com>
Date:   Wed, 16 Aug 2023 01:58:27 +0900
From:   Sergey Senozhatsky <senozhatsky@...omium.org>
To:     Robin Murphy <robin.murphy@....com>
Cc:     Christoph Hellwig <hch@....de>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Petr Mladek <pmladek@...e.com>,
        Rob Clark <robdclark@...omium.org>,
        John Ogness <john.ogness@...utronix.de>,
        linux-kernel@...r.kernel.org,
        Sergey Senozhatsky <senozhatsky@...omium.org>
Subject: Re: [PATCH] dma-debug: defer __dma_entry_alloc_check_leak() printk
 output

On (23/08/16 01:52), Sergey Senozhatsky wrote:
> On (23/08/15 17:42), Robin Murphy wrote:
> > On 15/08/2023 4:26 pm, Sergey Senozhatsky wrote:
> > > __dma_entry_alloc_check_leak() calls printk -> serial console
> > > output (qcom geni) and grabs port->lock under free_entries_lock,
> > > which is a conflicting locking dependency chain as qcom_geni IRQ
> > > handler can call into dma-debug code and grab free_entries_lock
> > > under port->lock.
> > > 
> > > Use deferred printk in __dma_entry_alloc_check_leak() so that we
> > > don't acquire serial console's port->lock under free_entries_lock.
> > 
> > Hmm, the print really doesn't need to be under the lock anyway, it only
> > needs to key off whether the "num_free_entries == 0" condition was hit or
> > not.
> 
> I thought about it, briefly. __dma_entry_alloc_check_leak() reads
> global nr_total_entries /  nr_prealloc_entries which are updated
> (inc/dec) under free_entries_lock, so I didn't want to move
> __dma_entry_alloc_check_leak() outside of free_entries_lock scope.

Something like this?

---

diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c
index 9e11ceadc69d..ca0508de4e78 100644
--- a/kernel/dma/debug.c
+++ b/kernel/dma/debug.c
@@ -637,15 +637,15 @@ static struct dma_debug_entry *__dma_entry_alloc(void)
        return entry;
 }

-static void __dma_entry_alloc_check_leak(void)
+static void __dma_entry_alloc_check_leak(u32 total_entries)
 {
-       u32 tmp = nr_total_entries % nr_prealloc_entries;
+       u32 tmp = total_entries % nr_prealloc_entries;

        /* Shout each time we tick over some multiple of the initial pool */
        if (tmp < DMA_DEBUG_DYNAMIC_ENTRIES) {
-               printk_deferred(KERN_INFO "dma_debug_entry pool grown to %u (%u00%%)\n",
-                               nr_total_entries,
-                               (nr_total_entries / nr_prealloc_entries));
+               pr_info("dma_debug_entry pool grown to %u (%u00%%)\n",
+                       total_entries,
+                       (total_entries / nr_prealloc_entries));
        }
 }

@@ -658,6 +658,8 @@ static struct dma_debug_entry *dma_entry_alloc(void)
 {
        struct dma_debug_entry *entry;
        unsigned long flags;
+       bool alloc_check_leak = false;
+       u32 total_entries;

        spin_lock_irqsave(&free_entries_lock, flags);
        if (num_free_entries == 0) {
@@ -667,13 +669,17 @@ static struct dma_debug_entry *dma_entry_alloc(void)
                        pr_err("debugging out of memory - disabling\n");
                        return NULL;
                }
-               __dma_entry_alloc_check_leak();
+               alloc_check_leak = true;
+               total_entries = nr_total_entries;
        }

        entry = __dma_entry_alloc();

        spin_unlock_irqrestore(&free_entries_lock, flags);

+       if (alloc_check_leak)
+               __dma_entry_alloc_check_leak(total_entries);
+
 #ifdef CONFIG_STACKTRACE
        entry->stack_len = stack_trace_save(entry->stack_entries,
                                            ARRAY_SIZE(entry->stack_entries),

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ