I think this has the wrong fixes tag, as pointed out by willy. It also does not fix the underlying failure of find_vma() not doing what is expected in nommu context. * Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> [230919 16:22]: > > The quilt patch titled > Subject: proc: nommu: fix empty /proc/<pid>/maps > has been removed from the -mm tree. Its filename was > proc-nommu-fix-empty-proc-pid-maps.patch > > This patch was dropped because it was merged into the mm-hotfixes-stable branch > of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm > > ------------------------------------------------------ > From: Ben Wolsieffer <ben.wolsieffer@xxxxxxxxxxx> > Subject: proc: nommu: fix empty /proc/<pid>/maps > Date: Fri, 15 Sep 2023 12:00:56 -0400 > > On no-MMU, /proc/<pid>/maps reads as an empty file. This happens because > find_vma(mm, 0) always returns NULL (assuming no vma actually contains the > zero address, which is normally the case). > > To fix this bug and improve the maintainability in the future, this patch > makes the no-MMU implementation as similar as possible to the MMU > implementation. > > The only remaining differences are the lack of hold/release_task_mempolicy > and the extra code to shoehorn the gate vma into the iterator. > > This has been tested on top of 6.5.3 on an STM32F746. > > Link: https://lkml.kernel.org/r/20230915160055.971059-2-ben.wolsieffer@xxxxxxxxxxx > Fixes: 0c563f148043 ("proc: remove VMA rbtree use from nommu") > Signed-off-by: Ben Wolsieffer <ben.wolsieffer@xxxxxxxxxxx> > Cc: Davidlohr Bueso <dave@xxxxxxxxxxxx> > Cc: Giulio Benetti <giulio.benetti@xxxxxxxxxxxxxxxxxxxxxx> > Cc: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx> > Cc: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx> > Cc: Oleg Nesterov <oleg@xxxxxxxxxx> > Cc: Vlastimil Babka <vbabka@xxxxxxx> > Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > --- > > fs/proc/internal.h | 2 -- > fs/proc/task_nommu.c | 37 ++++++++++++++++++++++--------------- > 2 files changed, 22 insertions(+), 17 deletions(-) > > --- a/fs/proc/internal.h~proc-nommu-fix-empty-proc-pid-maps > +++ a/fs/proc/internal.h > @@ -289,9 +289,7 @@ struct proc_maps_private { > struct inode *inode; > struct task_struct *task; > struct mm_struct *mm; > -#ifdef CONFIG_MMU > struct vma_iterator iter; > -#endif > #ifdef CONFIG_NUMA > struct mempolicy *task_mempolicy; > #endif > --- a/fs/proc/task_nommu.c~proc-nommu-fix-empty-proc-pid-maps > +++ a/fs/proc/task_nommu.c > @@ -175,15 +175,28 @@ static int show_map(struct seq_file *m, > return nommu_vma_show(m, _p); > } > > -static void *m_start(struct seq_file *m, loff_t *pos) > +static struct vm_area_struct *proc_get_vma(struct proc_maps_private *priv, > + loff_t *ppos) > +{ > + struct vm_area_struct *vma = vma_next(&priv->iter); > + > + if (vma) { > + *ppos = vma->vm_start; > + } else { > + *ppos = -1UL; > + } > + > + return vma; > +} > + > +static void *m_start(struct seq_file *m, loff_t *ppos) > { > struct proc_maps_private *priv = m->private; > + unsigned long last_addr = *ppos; > struct mm_struct *mm; > - struct vm_area_struct *vma; > - unsigned long addr = *pos; > > - /* See m_next(). Zero at the start or after lseek. */ > - if (addr == -1UL) > + /* See proc_get_vma(). Zero at the start or after lseek. */ > + if (last_addr == -1UL) > return NULL; > > /* pin the task and mm whilst we play with them */ > @@ -205,12 +218,9 @@ static void *m_start(struct seq_file *m, > return ERR_PTR(-EINTR); > } > > - /* start the next element from addr */ > - vma = find_vma(mm, addr); > - if (vma) > - return vma; > + vma_iter_init(&priv->iter, mm, last_addr); > > - return NULL; > + return proc_get_vma(priv, ppos); > } > > static void m_stop(struct seq_file *m, void *v) > @@ -227,12 +237,9 @@ static void m_stop(struct seq_file *m, v > priv->task = NULL; > } > > -static void *m_next(struct seq_file *m, void *_p, loff_t *pos) > +static void *m_next(struct seq_file *m, void *_p, loff_t *ppos) > { > - struct vm_area_struct *vma = _p; > - > - *pos = vma->vm_end; > - return find_vma(vma->vm_mm, vma->vm_end); > + return proc_get_vma(m->private, ppos); > } > > static const struct seq_operations proc_pid_maps_ops = { > _ > > Patches currently in -mm which might be from ben.wolsieffer@xxxxxxxxxxx are > >
