Fix POSIX skin build for 2.4.x
PATCH 1/1 Fix POSIX skin build for 2.4.x
Hi Gilles,
the backport from trunk was borken. This patch makes 2.4.x build again, but - BIG-FAT-WARNING - I haven't tested the result.
Jan
Signature · 2 lines
Siemens AG, Corporate Technology, CT SE 2 Corporate Competence Center Embedded Linux
ksrc/skins/posix/syscall.c +8 -6
@@ -1847,7 +1847,7 @@ static int __mq_send(struct task_structif (IS_ERR (msg))return PTR_ERR (msg); -if (__xn_copy_from_user (msg->data, +if (__xn_copy_from_user (curr, msg->data, (void __user *)__xn_reg_arg2 (regs), len)) {pse51_mq_finish_send (ufd->kfd, mq, msg);return -EFAULT ;
@@ -1903,7 +1903,7 @@ static int __mq_timedsend(struct task_stif (IS_ERR (msg))return PTR_ERR (msg); -if (__xn_copy_from_user (msg->data, +if (__xn_copy_from_user (curr, msg->data, (void __user *)__xn_reg_arg2 (regs), len)) {pse51_mq_finish_send (ufd->kfd, mq, msg);return -EFAULT ;
@@ -1952,7 +1952,7 @@ static int __mq_receive(struct task_struif (IS_ERR (msg))return PTR_ERR (msg); -if (__xn_copy_to_user ((void __user *)__xn_reg_arg2 (regs), +if (__xn_copy_to_user (curr, (void __user *)__xn_reg_arg2 (regs), msg->data, msg->len)) {pse51_mq_finish_rcv (ufd->kfd, mq, msg);return -EFAULT ;
@@ -2029,7 +2029,7 @@ static int __mq_timedreceive(struct taskif (IS_ERR (msg))return PTR_ERR (msg); -if (__xn_copy_to_user ((void __user *)__xn_reg_arg2 (regs), +if (__xn_copy_to_user (curr, (void __user *)__xn_reg_arg2 (regs), msg->data, msg->len)) {pse51_mq_finish_rcv (ufd->kfd, mq, msg);return -EFAULT ;
@@ -2041,8 +2041,10 @@ static int __mq_timedreceive(struct taskif (err)return err; -if (__xn_safe_copy_to_user ((void __user *)__xn_reg_arg3 (regs), - &len,sizeof (len))) +if (!__xn_access_ok + (curr,VERIFY_WRITE ,__xn_reg_arg3 (regs),sizeof (len)) || +__xn_copy_to_user (curr, (void __user *)__xn_reg_arg3 (regs), + &len,sizeof (len)))return -EFAULT ;if (__xn_reg_arg4 (regs))
Quoted 4 lines · Jan Kiszka
Hi Gilles,
the backport from trunk was borken. This patch makes 2.4.x build again, but - BIG-FAT-WARNING - I haven't tested the result.
I was almost sure I added the additional argument to xn_copy_from/to_user. It appears I did not. Anyway, I will apply your patch, thanks.
Signature · 1 line
Gilles Chanteperdrix