| Message ID | 20181218171834.GA14211@embeddedor (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | ALSA: rme9652: Fix potential Spectre v1 vulnerability | expand |
Hi,
[This is an automated email]
This commit has been processed because it contains a -stable tag.
The stable tag indicates that it's relevant for the following trees: all
The bot has tested the following trees: v4.19.10, v4.14.89, v4.9.146, v4.4.168, v3.18.130,
v4.19.10: Build OK!
v4.14.89: Build OK!
v4.9.146: Build OK!
v4.4.168: Build OK!
v3.18.130: Failed to apply! Possible dependencies:
6cbbfe1c8ddb ("ALSA: Include linux/io.h instead of asm/io.h")
How should we proceed with this patch?
--
Thanks,
Sasha
On Tue, 18 Dec 2018 18:18:34 +0100, Gustavo A. R. Silva wrote: > > info->channel is indirectly controlled by user-space, hence leading to > a potential exploitation of the Spectre variant 1 vulnerability. > > This issue was detected with the help of Smatch: > > sound/pci/rme9652/hdsp.c:4100 snd_hdsp_channel_info() warn: potential spectre issue 'hdsp->channel_map' [r] (local cap) > > Fix this by sanitizing info->channel before using it to index hdsp->channel_map > > Notice that given that speculation windows are large, the policy is > to kill the speculation on the first load and not worry if it can be > completed with a dependent load/store [1]. > > Also, notice that I refactored the code a bit in order to get rid of the > following checkpatch warning: > > ERROR: do not use assignment in if condition > FILE: sound/pci/rme9652/hdsp.c:4103: > if ((mapped_channel = hdsp->channel_map[info->channel]) < 0) > > > [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2 > > Cc: stable@vger.kernel.org > Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com> Applied, thanks. Takashi
diff --git a/sound/pci/rme9652/hdsp.c b/sound/pci/rme9652/hdsp.c index 1bff4b1b39cd..ba99ff0e93e0 100644 --- a/sound/pci/rme9652/hdsp.c +++ b/sound/pci/rme9652/hdsp.c @@ -30,6 +30,7 @@ #include <linux/math64.h> #include <linux/vmalloc.h> #include <linux/io.h> +#include <linux/nospec.h> #include <sound/core.h> #include <sound/control.h> @@ -4092,15 +4093,16 @@ static int snd_hdsp_channel_info(struct snd_pcm_substream *substream, struct snd_pcm_channel_info *info) { struct hdsp *hdsp = snd_pcm_substream_chip(substream); - int mapped_channel; + unsigned int channel = info->channel; - if (snd_BUG_ON(info->channel >= hdsp->max_channels)) + if (snd_BUG_ON(channel >= hdsp->max_channels)) return -EINVAL; + channel = array_index_nospec(channel, hdsp->max_channels); - if ((mapped_channel = hdsp->channel_map[info->channel]) < 0) + if (hdsp->channel_map[channel] < 0) return -EINVAL; - info->offset = mapped_channel * HDSP_CHANNEL_BUFFER_BYTES; + info->offset = hdsp->channel_map[channel] * HDSP_CHANNEL_BUFFER_BYTES; info->first = 0; info->step = 32; return 0;
info->channel is indirectly controlled by user-space, hence leading to a potential exploitation of the Spectre variant 1 vulnerability. This issue was detected with the help of Smatch: sound/pci/rme9652/hdsp.c:4100 snd_hdsp_channel_info() warn: potential spectre issue 'hdsp->channel_map' [r] (local cap) Fix this by sanitizing info->channel before using it to index hdsp->channel_map Notice that given that speculation windows are large, the policy is to kill the speculation on the first load and not worry if it can be completed with a dependent load/store [1]. Also, notice that I refactored the code a bit in order to get rid of the following checkpatch warning: ERROR: do not use assignment in if condition FILE: sound/pci/rme9652/hdsp.c:4103: if ((mapped_channel = hdsp->channel_map[info->channel]) < 0) [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2 Cc: stable@vger.kernel.org Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com> --- sound/pci/rme9652/hdsp.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)