karg.hdr.id and port are 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: drivers/message/fusion/mptctl.c:1360 mptctl_getiocinfo() warn: potential spectre issue 'ioc->pfacts' [r] drivers/message/fusion/mptctl.c:2788 mptctl_hp_targetinfo() warn: potential spectre issue 'hd->sel_timeout' [r] (local cap) Fix this by sanitizing both karg.hdr.id and port before using them to index ioc->pfacts and hd->sel_timeout. 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]. [1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2 Cc: sta...@vger.kernel.org Signed-off-by: Gustavo A. R. Silva <gust...@embeddedor.com> --- drivers/message/fusion/mptctl.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/message/fusion/mptctl.c b/drivers/message/fusion/mptctl.c index 8d22d6134a89..f6c1311ae86f 100644 --- a/drivers/message/fusion/mptctl.c +++ b/drivers/message/fusion/mptctl.c @@ -66,6 +66,8 @@ #include <scsi/scsi_host.h> #include <scsi/scsi_tcq.h> +#include <linux/nospec.h> + #define COPYRIGHT "Copyright (c) 1999-2008 LSI Corporation" #define MODULEAUTHOR "LSI Corporation" #include "mptbase.h" @@ -1306,7 +1308,7 @@ mptctl_getiocinfo (unsigned long arg, unsigned int data_size) kfree(karg); return -EINVAL; } - port = karg->hdr.port; + port = array_index_nospec(karg->hdr.port, 2); karg->port = port; pdev = (struct pci_dev *) ioc->pcidev; @@ -2689,6 +2691,7 @@ mptctl_hp_targetinfo(unsigned long arg) } if (karg.hdr.id >= MPT_MAX_FC_DEVICES) return -EINVAL; + karg.hdr.id = array_index_nospec(karg.hdr.id, MPT_MAX_FC_DEVICES); dctlprintk(ioc, printk(MYIOC_s_DEBUG_FMT "mptctl_hp_targetinfo called.\n", ioc->name)); -- 2.19.2
